Vortices generated when flow pass offshore slender structures may cause structural vibrations due to fluid-structure interaction, termed vortex-induced vibrations (VIV). VIV can cause severe accumulated fatigue damage, which need to be considered in the structure design. To capture the non-linearities of different riser systems and time varying flow, time domain models are better alternatives than frequency domain models to predict VIV. The input hydrodynamic parameters of time domain models are semi-empirical. However, versatile databases of hydrodynamic parameters are not well established, due to the large variations of the flow conditions, riser systems, and corresponding sub-structures. An optimization procedure could be implemented as an alternative to conventional experimental methods for establishing adaptive hydrodynamic parameter databases capable of accurately handling changes in the underlying structure of the system or flow regime. Literatures on application of optimization algorithms for various offshore engineering problems have been reviewed. VIV represents a constrained non-linear programming problem, where the sequential quadratic programming (SQP) algorithm – NLPQLP has been applied in this study. This paper proposes an optimization procedure of input hydrodynamic parameters for the time domain VIV solver - VIVANA-TD [21], for VIV prediction of various riser systems. A workflow integrating optimization and time domain VIV simulation has been established in the software work bench SIMA. By investigating the VIV prediction of typical laboratory model tests with optimized parameters, the feasibility of optimization using sequential quadratic non-linear programming algorithm - NLPQLP has been demonstrated.

In this paper we consider the allocation problem within the context of optimal motion control for floating or submerged bodies. The purpose of our research is to develop an allocation algorithm which allows for (1) multiple bodies with up to six modes for each body, (2) arbitrary actuator types — azimuthing thrusters, propeller-rudder systems etc., (3) arbitrary objective functions, (4) interaction effects such as forbidden zones, and (5) physical limitations such as saturation. Some ideas presented in [1, 2] were generalised to more widely applicable concepts. Each body has an arbitrary number of actuators, each actuator has an arbitrary number of degrees of freedom. Interaction effects are modelled by means of state-dependent effectivity coefficients. Coupled states, such as propeller thrust and torque, are modelled as linearised constraints. The constrained optimization problem is solved with a combination of Sequential Quadratic Programming and Steepest Descent methods. The Python implementation is coupled with MARIN’s extensible modelling framework (XMF). We demonstrate the generic allocation algorithm for an underwater vehicle with multiple actuator types, physical limitations and coupled states and for a surface vessel with two propeller-rudder systems and a bow tunnel thruster. The results show that the allocation algorithm is able to handle complex configurations with specific physical limitations and coupled modes while adopting a generic approach.

This paper describes the experimental assessment of the solution developed by IFP Energies nouvelles (IFPEN) for the WEC Control Competition (WECCCOMP), a benchmark devised to compare energy-maximizing controllers for wave energy converters (WECs). For the first round of the competition, carried out in simulation with WEC-Sim, IFPEN had submitted an MPC-based solution, which eventually scored best among the contestants, with respect to the energy-related criterion defined by the organizers and computed on a set of six irregular waves. For the second round of the competition, the performance of this MPC solution has been assessed in a tank test situation, on the same Wavestar-like scale device that had been simulated in WEC-Sim. The paper first recalls the features of the control solution implemented for the competition, an offline-tuned weighted-QP MPC algorithm using short-term prediction from present and past wave excitation force estimates. Then, the major steps of the test plan are described and control design choices are explained. Finally, the experimental results in closed loop are presented. Despite noise and an imperfectly controlled PTO actuator, IFPEN’s MPC solution proves efficient and robust: its experimental performance in terms of harvested electrical energy is in accordance with the nominal results obtained in simulation using the linear design model, and this, for different realizations of each sea state. This performance has allowed the IFPEN team to win the competition also at the experimental stage.

Economics decision drives the operation of ocean-wave energy converters (WEC) to be in a “farm mode”. Control strategy developed for a WEC array will be of high importance for improving the aggregate energy extraction efficiency of the whole system. Model-predictive control (MPC) has shown its strong potential in maximizing the energy output in devices with hard constraints on operation states and machinery inputs (See Ref. [1–3]). Computational demands for using MPC to control an array in real time can be prohibitive. In this paper, we formulate the MPC to control an array of heaving point absorbers, by recasting the optimization problem for energy extraction into a convex Quadratic Programming (QP) problem, the solution of which can be carried out very efficiently. Large slew rates are to be penalized, which can also guarantee the convexity of the QP and improve the computational efficiency for achieving the optimal solution. Constraints on both the states and the control input can be accommodated in this MPC method. Full hydro-dynamic interference effects among the WEC array components are taken into account using the theory developed in [4]. Demonstrative results of the application are presented for arrays of two, three, and four point absorbers operating at different incident-wave angles. Effects of the interacting waves on power performance of the array under the new MPC control are investigated, with simulations conducted in both regular and irregular seas. Heaving motions of individual devices at their optimal conditions are shown. Also presented is the reactive power required by the power takeoff (PTO) system of the array to achieve optimality. We are pleased to contribute this article in celebration of our collegiality with Professor Bernard Molin on the occasion of his honoring symposium.

Design approaches for wave impact on marine structures in storm sea-states are being reconsidered due to events related to the safety of North Sea offshore structures, both fixed and floating (Valhall QP extended lifetime; COSL Innovator accident). There has been a strong research and tool development within the field during the last decade, both within model testing and numerical analysis, including CFD. However, there is still a lack of efficient methods and tools to properly analyze these phenomena and their probably of occurrence. One major aspect in this is to reduce the statistical uncertainties that are naturally arising in estimates of design loads related to extreme waves. In order to estimate the design loads it is common practice not to investigate all possible sea states (i.e. long-term analysis) but to investigate a few sea states and assume that the design value occurs at a prescribed probability level in the sea states with the same probability level (i.e. contour line approach). The estimate of the design value at that probability level is then based on results from a limited number of random realizations of these sea states. For linear or weakly nonlinear response types it is possible to estimate design loads accurately with a quite limited number of realizations. For strongly nonlinear/badly behaved problems however this is not possible due to the large variations in the tail of the distribution of the impact load, and many more realizations are required. This means that much more of these extreme, rare impact-related events should be collected to reduce the statistical uncertainty in the design load. This approach is restricted by time and costs and eventually one may have to accept an estimated design load with a large statistical uncertainty and account for the uncertainty with a higher safety margin. In this paper an improved methodology for estimating design loads related to extreme wave impacts will be presented. The methodology is based on screening many 3-hour realizations of the design sea states with simplified, fast but sufficiently accurate methods and to focus only on the potentially critical events with a model containing a more complete description of the physics. This can be either a model test or a non-linear impact simulation (i.e. CFD analysis). By doing this many more rare/critical events can be assessed, reducing the statistical uncertainty in the estimate of the design load. The main challenge is to find suitable screening methods, which may be different for different structures (fixed, floating, etc.). Several screening methods/wave impact indicators will be discussed, and their capability will be illustrated by analyzing existing model test data for fixed and floating structures, showing the correlation between indicator and actual impact events.

Model-Predictive Control (MPC) has shown its strong potential in maximizing energy extraction for Wave-Energy Converters (WECs) while handling hard constraints. As MPC can solve the optimization problem on-line, it can better account for state changes and reject disturbances from the harsh sea environment. Interests have arisen in applying MPC to an array of WECs, since researchers found that multiple small-size WECs are more economically viable than a single large-size WEC. However, the computational demand is known to be a primary concern for applying MPC in real-time, which can determine the feasibility of such a controller, particularly when it comes to controlling an array of absorbers. In this paper, we construct a cost function and cast the problem into a Quadratic Programming (QP) with the machinery force being the “optimizer,” for which the convexity can be guaranteed by introducing a penalty term on the slew rate of the machinery force. The optimization problem can then be solved efficiently, and a feasible solution will be assured as the global optima. Constraints on the motion of the WEC and the machinery force will be taken into account. The current MPC will be compared to others existing in literature, including a nonlinear MPC [1] which has been applied in wave-tank tests. The effects of constraints on the control law and the absorbed power are investigated. Performances of the WEC are shown for both regular and irregular wave conditions. The current MPC is found to have good energy-capture capability and is able to broaden the band-width for capturing wave energy. The reactive power required by the PTO system is presented. The additional penalty term provides a tuning parameter, of which the effects on the MPC performance and the reactive power requirement are discussed.

A method was recently presented for determining quasi-static and dynamic riser angles using measured data typically found in a riser fatigue monitoring system, specifically acceleration and angular rate data. The riser angles were determined at sensor locations. In this paper quasi-static riser displacement, inclination angle, curvature, and stress are estimated along the entire length of the riser, using only the quasi-static inclinations angles at sparse sensor locations. In addition the distribution of applied forces along the entire riser length is also estimated. A rough representation of the current profile can be calculated using the drag coefficients of riser joints. The riser deformation (displacement, inclination, curvature) and applied forces are estimated by solving the matrix equation f = K * x , where f is the vector of forces and moments, K is the stiffness matrix and x is the vector of displacements and inclination angles. In the equation, force and displacement vectors are unknown and the stiffness matrix is determined using Finite Element (FE) modeling. Constraints are applied, setting the inclination angle at the sensor locations to the values derived from measured data. The remaining highly-underdetermined problem cannot be solved in a classical sense, as it admits infinite solutions. To get a solution that is consistent with the physics of riser deformation, smoothness of the solution is enforced as a constraint. The smoothest solution is solved using quadratic programming methods. Following implementation of the method in Matlab ® , the procedure was validated using numerical simulations of a riser in applied current. Both connected (to the wellhead) and disconnected cases were simulated. Estimated riser displacements, slopes, curvatures and applied forces are found to match the simulation results closely. The algorithm was then run using measured data from an emergency disconnect event that occurred on the Chikyu drill ship in November, 2012. The riser displacement, inclination and curvature were determined and found to agree well with results determined using another method. The additional capabilities presented herein further expand the utility of a riser monitoring system. Quasi-static and dynamic riser angles are derived from acceleration and angular rate sensors using previously published methods. Using the method developed herein, the quasi-static inclination angles at the sensor locations can be used to determine the displacement, inclination, curvature (stress) and even applied force along the entire riser. These results are particularly useful in strength assessment, model verification, clashing and emergency event reconstruction analyses.

The Dynamic Positioning (DP) system is responsible for the stationkeeping and manoeuvrability of a vessel in offshore operations. The forces required by the DP system are distributed among the available thrusters of the vessel by a thrust allocation algorithm, which should be both accurate — i.e. the sum of the effective forces is close to the required forces — and efficient — i.e. the total consumed power is minimal. Ideally, the thrust allocation algorithm accounts for the forbidden zones, a practical solution to avoid significant hydrodynamic interaction effects such as thruster-hull interaction, thruster-current interaction and thruster-thruster interaction. Moreover, the thrust allocation algorithm should also take into account the physical limitations of each thruster such as the maximum thrust (saturation), the maximum rate of turn (azimuth) and the maximum rate of change of RPM. In order to include these complex requirements in the thrust allocation algorithm the hydrodynamic interaction effects are modelled as efficiency functions, which are incorporated in the (power) object function and in the (required forces) constraints. With the purpose of solving the complete problem, including the physical limitations and the hydrodynamic interaction effects, several advanced optimization methods were investigated. The selected optimization algorithms were the Sequential Quadratic Programming (SQP) method and the Steepest Descent (SD) method. In this paper a DP drill ship is considered, with 6 azimuthing thrusters. Results were obtained for environmental conditions of increasing strength, from benign sea states to higher sea states with large required forces, eventually leading to thruster saturation. As a final test case, the required forces are such that the thrust allocation algorithm forces the thrusters to cross large forbidden zones with significant thruster-hull interaction. The results show that the enhanced thrust allocation algorithm can deal with both hydrodynamic interaction effects and physical limitations in a time-efficient and robust manner. Altogether, these improvements are expected to lead to higher accuracy and efficiency for DP operations.

In this work, the implementation of reliability-based optimization (RBO) of a circular steel monopod-offshore-tower with constant and variable diameters (represented by segmentations) and thicknesses is presented. The tower is subjected to the extreme wave loading. For this purpose, the deterministic optimization of the tower is performed with constraints including stress, buckling, and the lowest natural frequency firstly. Then, a reliability-based optimization of the tower is performed. The reliability index is calculated from FORM using a limit state function based on the lowest natural frequency. The mass of the tower is considered as being the objective function; the thickness and diameter of the cross-section of the tower are taken as being design variables of the optimization. The numerical strategy employed for performing the optimization uses the IMSL-Libraries routine that is based on the Sequential Quadratic Programming (SQP). In addition, to check the results obtained from aforementioned procedure, the RBO of the tower is also performed using the genetic algorithms (GA) tool of the MATLAB. Finally, a demonstration of an example monopod tower is presented.

The paper presents improved methods and new results on the introduction of formal optimization strategies into the design of offshore structures. The hull design stage is singled out from the overall design process and automated by introducing parametric shape generation, numeric hydrodynamic analysis and assessment tools as well as Nonlinear Programming algorithms for process control. The investigation compares the performance of three different optimization algorithms within a shape optimization framework. The classical deterministic Sequential Quadratic Programming method competes with two so called global optimization algorithms: The popular Genetic Algorithm and the more exotic Adaptive Simulated Annealing. The applications show that significant improvements of seakeeping qualities are obtained in either case. As expected, the global methods require definitely more computation time than the deterministic algorithm. Furthermore the global methods do not always produce better results, which makes a careful choice of the optimization algorithm mandatory. Guidelines for an efficient application are given in the conclusions.