Modelling and geometry optimisation of wave energy
Transcription
Modelling and geometry optimisation of wave energy
Modelling and geometry optimisation of wave energy converters Adi Kurniawan Supervisors: Prof. Torgeir Moan & Prof. em. Johannes Falnes 23 April 2013 Outline Introduction Modelling Geometry optimisation Geometry control Conclusion There is abundant power available from the waves NatGeo TV, “Green DIY Riding radical wave power” Any device will deliver some energy What matters is the cost of energy Ultimate problem Given the waves, design a device that minimises the cost per unit of delivered energy. What matters is the cost of energy Ultimate problem Given the waves, design a device that minimises the cost per unit of delivered energy. Open questions How does this device look like? Is there a systematic way to design an economic device? What do we need? What do we need? a model What do we need? a model predict power output & device response What do we need? a model predict power output & device response an optimiser What do we need? a model predict power output & device response maximise energy & minimise cost an optimiser What do we need? a model predict power output & device response maximise energy & minimise cost an optimiser Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Assumptions I Linear hydrodynamics I Unidirectional incident wave I Uniform water depth Outline Introduction Modelling Geometry optimisation Geometry control Conclusion What is a WEC? An oscillator in water What is a WEC? It has two main parts What is a WEC? primary interface It has two main parts What is a WEC? power take-off (PTO) primary interface It has two main parts Oscillating-body WECs Utilise relative motion I between a moving body and a fixed reference (e.g. sea bed) Oscillating-body WECs Utilise relative motion I between a moving body and a fixed reference (e.g. sea bed) I between several moving bodies Oscillating-water-column WECs Utilise the motion of a partially enclosed mass of water relative to I a fixed reference I a moving outer body A WEC is faced with two inherent challenges I A WEC must work with large forces and low velocities I I Current technology for electrical generation is more used to low force, high speed motions Large forces are inconvenient and expensive to handle A WEC is faced with two inherent challenges I A WEC must work with large forces and low velocities I I I Current technology for electrical generation is more used to low force, high speed motions Large forces are inconvenient and expensive to handle A WEC must deal with the stochastic nature of the waves I I It has to absorb energy optimally from the most frequent waves It has to survive the most extreme waves Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Several energy conversion alternatives are available Mechanical translation/rotation Hydraulic piston Mechanical system Hydraulic Hydraulic Pneumatic Water turbine Hydraulic motor Air turbine Mechanical rotation Electrical generator Electrical Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling Bond graph modelling Domain-independent; uses common notations across different energy domains I Se 1 R C Bond graph modelling Port-based, modular I Se 1 R C Bond graph modelling Port-based, modular I Se 1 C R Se I C I 1 0 1 R R C Bond graph modelling Graphical and intuitive 20-sim Reference 4.3 Bond graph modelling Systematic derivation of model equations → automated simulation Bond graph modelling Systematic derivation of model equations → automated simulation Bond graph modelling in wave energy research First allusion by Jefferys (1984), “Simulation of wave power devices” Bond graph modelling in wave energy research Otherwise, relatively recent Nolan et al. (2003) PTO of a WEC for electricity and potable water production Marré (2006) WEC consisting of a buoy connected to a semi-submersible Engja and Hals (2007) WEC consisting of a buoy connected to a semi-submersible Bacelli et al. (2008) hydraulic PTO Hals (2010) integrated WEC systems Yang (2011) hydraulic PTO Single oscillating body Hydrost. Restoring Exc. Force Mass + m(∞) C I 1 Se Coulomb Damping R R Quadratic Damping TF Transformation Ext. Restoring C R Radiation Impedance 0 1 P R Load Resistance Self-reacting bodies Hydrost. Restoring C Se Mass + m(∞) I Quadratic Damping R 1 Coulomb Damping Exc. Force Ext. Restoring R R Radiation Impedance C Se 1 Exc. Force C Hydrost. Restoring I Mass + m(∞) R Quadratic Damping 0 1 P R Load Resistance Self-reacting bodies Hydrost. Restoring C Se Mass + m(∞) I Quadratic Damping R 1 Coulomb Damping Exc. Force Ext. Restoring R R Radiation Impedance C Se 1 C 1 P Mass + m(∞) R I Load Resistance 1 Exc. Force I Mass + m(∞) R Quadratic Damping R Rad. Impedance + Ext. Damping Se Exc. Force Hydrost. Restoring 0 R Load Resistance C Hyd. + Ext. Restoring Oscillating water column (OWC) The hydrodynamics of an OWC device may be modelled using I massless piston model Oscillating water column (OWC) The hydrodynamics of an OWC device may be modelled using I massless piston model → oscillating-body problem, approximate Oscillating water column (OWC) The hydrodynamics of an OWC device may be modelled using I massless piston model → oscillating-body problem, approximate I pressure distribution model Oscillating water column (OWC) The hydrodynamics of an OWC device may be modelled using I massless piston model → oscillating-body problem, approximate I pressure distribution model → accurate Floating OWC Floating OWC Fj = fj A − X Zjj 0 Uj 0 − Hjp p, j0 Q = qA − Y p − X j = 1, 2, . . . , 6 HjU Uj j fj q Zjj 0 Y Hjp , HjU excitation force coefficient excitation vol. flow coefficient radiation impedance radiation admittance radiation coupling coefficients Three options are available to evaluate Hjp and Y 1. Modify the dynamic boundary condition on the internal free surface by special coding Three options are available to evaluate Hjp and Y 1. Modify the dynamic boundary condition on the internal free surface by special coding 2. Use generalised modes Three options are available to evaluate Hjp and Y 1. Modify the dynamic boundary condition on the internal free surface by special coding 2. Use generalised modes 3. Use reciprocity relations G = <{Y } = k 8πρgvg Z π −π Hjp = −HjU |q(β)|2 dβ Fixed & floating OWC C 0 1 R Sf Air Compressibility P R Load Resistance Relief Valve 0 Exc. Vol. Flow R Ext. Damping R Radiation Admittance Fixed OWC Fixed & floating OWC Hydrost. Restoring Exc. Force Mass + m(∞) C I 1 Se Coulomb Damping R R Quadratic Damping TF Transformation Ext. Restoring C R Radiation Impedance 0 1 P R Load Resistance Oscillating body Fixed & floating OWC Hydrost. Restoring Exc. Force Mass + m(∞) I C Se 1 Coulomb Damping R R Quadratic Damping TF Transformation Ext. Restoring C C R Radiation + C(∞) Coupling TF Radiation Impedance 0 1 R Sf Air Compressibility P R Load Resistance Relief Valve 0 Exc. Vol. Flow R Ext. Damping R Radiation Admittance Floating OWC To be realistic, time-domain model is necessary Assuming PTO force as linear enables frequency-domain analysis. Fe (ω) = R(ω) + Ru + iω M + Mu + m(ω) − (Sb + Su )ω −2 U (ω) However, since the PTO is nonlinear in reality, time-domain model is necessary. Fe (t) = [M + m(∞)]u̇(t) + k(t) ∗ u(t) + Sb s(t) + Fext (s(t), u(t), t) Fe s(t) U (ω), u(t) M m(ω) m(∞) R(ω) k(t) Sb Ru , M u , S u Fext (s(t), u(t), t) wave excitation force body displacement body velocity structural inertia added inertia infinite-frequency added inertia radiation damping radiation impedance impulse response function (IRF) hydrostatic stiffness PTO damping, inertia, and spring general nonlinear force which includes the PTO force To be realistic, time-domain model is necessary Assuming PTO force as linear enables frequency-domain analysis. Fe (ω) = R(ω) + Ru + iω M + Mu + m(ω) − (Sb + Su )ω −2 U (ω) However, since the PTO is nonlinear in reality, time-domain model is necessary. Fe (t) = [M + m(∞)]u̇(t) + k(t) ∗ u(t) + Sb s(t) + Fext (s(t), u(t), t) Fe s(t) U (ω), u(t) M m(ω) m(∞) R(ω) k(t) Sb Ru , M u , S u Fext (s(t), u(t), t) wave excitation force body displacement body velocity structural inertia added inertia infinite-frequency added inertia radiation damping radiation impedance impulse response function (IRF) hydrostatic stiffness PTO damping, inertia, and spring general nonlinear force which includes the PTO force Force equations for a floating OWC Frequency domain: Fj = fj A − X Zjj 0 Uj 0 −Hjp p, j0 Q = qA − Y p − X j = 1, 2, . . . , 6 HjU Uj j Time domain: Fj (t) = Fe,j (t) − X [mjj 0 (∞)u̇j 0 (t) + kjj 0 (t) ∗ uj 0 (t)] j0 +Cj (∞)p(t) + hj (t) ∗ p(t), Q(t) = Qe (t) − y(t) ∗ p(t) − X j j = 1, 2, . . . , 6 [Cj (∞)uj (t) + hj (t) ∗ uj (t)] The convolution accounts for the system memory Rt The convolution k(t) ∗ u(t) ≡ 0 k(t − τ )u(τ ) dτ accounts for the system memory, signifying the fact that waves radiated by the body in the past continue to affect the body for all subsequent times. Evaluation of the convolution term is problematic It has to be re-evaluated at every time step. Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, assessment of alternatives to the radiation convolution term Alternatives are available to avoid the convolution term I State-space model A set of coupled linear ordinary differential equations I Constant-coefficient model Replace frequency-dependent radiation coefficients by constants Three generic WECs I Single oscillating body I Fixed OWC I Floating OWC Various nonlinearities are included in the models I Quadratic damping I Coulomb damping osc. body Various nonlinearities are included in the models I Quadratic damping I Coulomb damping I Air compressibility I Pressure relief valve osc. body fixed OWC Various nonlinearities are included in the models I Quadratic damping I Coulomb damping I Air compressibility I Pressure relief valve osc. body fixed OWC floating OWC State-space models are found to be accurate error [%] 6 SS2 SS3 SS4 SS7 4 2 0 0 2 ω [rad/s] 4 6 Errors relative to frequency-domain model, for the body velocity of a WEC without nonlinear terms computation time [s] State-space models save computation time 600 ode4 SS3 400 200 0 0 2000 4000 6000 8000 10000 simulation length [s] Computation time of state-space radiation model of order 3 (SS3) compared to direct convolution integration (ode4) Application 1: Oscillating body with hydraulic PTO Width Draft 10 7.5 m m Two alternative hydraulic PTO systems HP_accumulator fluid_inertance C I Se Exciting_force Body TF P Primary conversion: Hydraulic cylinder 0 1 Check valves 0 TF R pipe_resistance C LP_accumulator Secondary conversion: Hydraulic motor P R R Two alternative hydraulic PTO systems pipe_resistance1 fluid_inertance1 I R Se Exciting_force Body TF P C 1 Primary conversion: Hydraulic cylinder HP_accumulator 0 Check valves 0 1 TF R pipe_resistance2 I fluid_inertance2 C LP_accumulator Secondary conversion: Hydraulic motor P R R Simulation: 2-valve system, Tp = 11 s, Hs = 2 m 6 x 10 0 F exc [Nm] 5 −5 η [rad] 0.5 5 0 −0.5 7 [Pa] 3.4 x 10 p HP 3.2 3 6 [Pa] 12 x 10 p LP 10 8 u P [kW] 150 100 50 0 0 100 200 300 time [s] 400 500 600 Simulation: 4-valve system, Tp = 11 s, Hs = 2 m 6 x 10 0 F exc [Nm] 5 −5 η [rad] 0.5 5 0 −0.5 7 [Pa] 3.4 x 10 p HP 3.2 3 6 [Pa] 12 x 10 p LP 10 8 u P [kW] 150 100 50 0 0 100 200 300 time [s] 400 500 600 Application 2: Backward bent duct buoy (BBDB) Application 2: Backward bent duct buoy (BBDB) Application 2: Backward bent duct buoy (BBDB) Length Width Draft 50 24 13 m m m Bond graph model of the BBDB Incident Wave Ampl. Hydrost. + Ext. Mass + m(∞) Ext. Damping Restoring C file input MSe I 1 Exc. Force R TF R body TF TF Air Compressibility C Radiation Impedance Radiation Coupling + C(∞) 0 R P Air Turbine Sf Radiation Admittance R Relief Valve MSf 0 Exc. Vol. Flow internal surface R Ext. Damping Mass + m(∞) With state-space models Hydrost. + Ext. Restoring I C Exc. Force Data f ile input Exc. Force 1 MSe SSRadImp Ext. Damping R body TF TF MR Rad. Imp. Air Comp. MSe SSRadCF C Rad. Coupl. F TF C(∞) 0 Air. Turb. MSf SSRadCQ R P Rad. Coupl. Q R Relief Valve Rad. Adm. AB CD MSf y f ile input Exc. Vol. Data MSf 0 Exc. Vol. Flow internal surface R Ext. Damping Simulation: Tp = 8 s, Hs = 3 m 1200 P (t), unlimited p u mean Pu, unlimited p Pu(t), limited p 800 mean Pu, limited p 600 u P [kW] 1000 400 200 0 0 50 100 150 t [s] 200 250 300 Outline Introduction Modelling Geometry optimisation Geometry control Conclusion A good wave absorber has to be a good wave maker |H̄r opt (β ± π)|2 Pmax = λJ R 2π 2 0 |Hr opt (θ)| dθ λ J Hr (β) β incident wavelength wave-power level radiation Kochin function wave heading angle To maximise power absorption, a system of WECs has to radiate waves mainly opposite the incident wave direction, when the system is forced to oscillate in time-reversal of its optimum motion, in otherwise calm water. A good wave absorber has to be a good wave maker Jamie Taylor (1976) But at what cost? ... a most important and urgent challenge is to develop a feasible single unit of a WEC, a unit that maximises the power output, not with respect to the free wave power that is available in the ocean, but with respect to parameters related more directly to the WEC itself (size, cost of investment and maintenance, etc.). Falnes and Hals (2012) We need to maximise energy absorption and minimise cost Maximise energy absorption I maximise power absorption I minimise losses I maximise design life I maximise capacity factor Minimise cost I minimise capital expenditures I minimise operational expenses I minimise environmental impact We need to maximise energy absorption and minimise cost Maximise energy absorption I maximise power absorption I minimise losses I maximise design life I maximise capacity factor Minimise cost I minimise capital expenditures I minimise operational expenses I minimise environmental impact conflicting objectives! Example: Designing a bridge The bridge is to carry as much load as possible and be as lightweight as possible → conflicting objectives! Wolfram (2007) Example: Designing a bridge The bridge is to carry as much load as possible and be as lightweight as possible → conflicting objectives! Wolfram (2007) Possible solution strategies I Combine multiple objectives into one aggregate objective function, e.g. relative capture width = capture width device dimension Possible solution strategies I Combine multiple objectives into one aggregate objective function, e.g. relative capture width = capture width device dimension Drawbacks: I I Need to know beforehand the relative importance of the objectives Returns only one optimum solution Possible solution strategies I Combine multiple objectives into one aggregate objective function, e.g. relative capture width = capture width device dimension Drawbacks: I I I Need to know beforehand the relative importance of the objectives Returns only one optimum solution Multi-objective optimisation algorithms I I Returns multiple optimum solutions to choose from Provides insight into the behaviour of the solutions along the optimal front Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Multi-objective optimisation The algorithm The algorithm The algorithm The algorithm The algorithm The algorithm The algorithm The algorithm The algorithm The algorithm Application 1: Composite circular cylinder Optimisation objectives: a2 θ c 1. maximise absorbed power, integrated over a wave frequency range a1 h d1 d/2 Variable d d1 a1 a2 min [m] max [m] 4 1 2 1 20 d/2 − 1 7 0.95 c Pmax = |Fe |2 , 4(R + |Zi |) where Zi = R+iω M + m − Sb ω −2 2. minimise total surface area → indicative of structural cost Progression of the ‘best’ solutions at the end of each generation 1000 As 800 × initial population 600 -◦- temporary Pareto fronts 400 200 0 0 -◦- final Pareto front 500 108 / 1500 R1000 Pmax dω 2000 Optimum geometries −10 0 5 10 15 x [m] 6 z [m] 0 −5 −10 0 5 10 15 x [m] 0 0 5 10 −5 15 600 400 200 0 0 0 5 −5 0 5 10 15 0 5 x [m] 2 5 500 7 108 / 8 9 10 1500 R1000 Pmax dω 8 z [m] 5 −5 0 5 10 15 x [m] 0 −5 −10 −5 1 6 15 0 x [m] 3 4 10 x [m] 2000 10 15 5 0 −5 −10 5 −5 −10 −5 x [m] 800 7 −5 −10 5 −5 −10 −5 As z [m] −10 5 −5 z [m] 5 −5 5 4 0 9 z [m] −10 −5 5 3 0 z [m] z [m] z [m] −5 5 2 0 z [m] 5 1 0 z [m] 5 −5 0 5 10 15 10 15 x [m] 0 −5 −10 −5 0 5 x [m] 10 15 −5 0 5 x [m] 10 Optimum geometries −10 0 5 10 15 x [m] 6 z [m] 0 −5 −10 0 5 10 15 x [m] 0 0 5 10 −5 15 600 400 200 0 0 0 5 10 15 x [m] 0 0 −5 5 10 15 z [m] 5 −5 0 5 10 15 x [m] 0 −5 −10 −5 x [m] 8 0 5 x [m] 10 15 5 0 −5 −10 5 −5 −10 −5 x [m] 800 7 −5 −10 5 −5 −10 −5 As z [m] −10 5 −5 z [m] 5 −5 5 4 0 9 z [m] −10 −5 5 3 0 z [m] z [m] z [m] −5 5 2 0 z [m] 5 1 0 z [m] 5 −5 0 5 10 15 10 15 x [m] 0 10 −5 −10 −5 0 5 x [m] 10 15 −5 0 5 x [m] I Among the optimum geometries, 1 more power can be absorbed only by increasing the surface area. 2 3 4 5 6 500 7 108 / 8 9 10 1500 R1000 Pmax dω 2000 Optimum geometries −10 0 5 10 15 x [m] 6 z [m] 0 −5 −10 0 5 10 15 x [m] 0 0 5 10 −5 15 600 400 200 0 0 0 5 10 15 x [m] 0 0 −5 5 10 15 z [m] 5 −5 0 5 10 15 x [m] 0 −5 −10 −5 x [m] 8 0 5 x [m] 10 15 5 0 −5 −10 5 −5 −10 −5 x [m] 800 7 −5 −10 5 −5 −10 −5 As z [m] −10 5 −5 z [m] 5 −5 5 4 0 9 z [m] −10 −5 5 3 0 z [m] z [m] z [m] −5 5 2 0 z [m] 5 1 0 z [m] 5 −5 0 5 10 15 10 15 x [m] 0 10 −5 −10 −5 0 5 x [m] 10 15 −5 0 5 x [m] I Among the optimum geometries, 1 more power can be absorbed only by increasing the surface area. 2 3 4 I The radii of the central cylinder 5 6 500 tend to the maximum limit. 7 108 / 8 9 10 1500 R1000 Pmax dω 2000 Optimum geometries −10 0 5 10 15 x [m] 6 z [m] 0 −5 −10 0 5 10 15 x [m] 0 0 5 10 −5 15 600 400 200 0 0 0 5 10 15 x [m] 0 0 −5 5 10 15 z [m] 5 −5 0 5 10 15 x [m] 0 −5 −10 −5 x [m] 8 0 5 x [m] 10 15 5 0 −5 −10 5 −5 −10 −5 x [m] 800 7 −5 −10 5 −5 −10 −5 As z [m] −10 5 −5 z [m] 5 −5 5 4 0 9 z [m] −10 −5 5 3 0 z [m] z [m] z [m] −5 5 2 0 z [m] 5 1 0 z [m] 5 −5 0 5 10 15 10 15 x [m] 0 10 −5 −10 −5 0 5 x [m] 10 15 −5 0 5 x [m] I Among the optimum geometries, 1 more power can be absorbed only by increasing the surface area. 2 3 4 I The radii of the central cylinder 5 6 tend to the maximum limit. 7 8 9 10 1500 R1000 108 / Pmax dω 500 I The radii of the larger cylinders 2000 are not maximized. For geometries 6 to 10, a1 = 2 m, the minimum limit. Added inertia, Sb ω −2 − M , and absorbed power 5 1 z [m] 0 −5 −10 −5 0 15 4000 5 4 3 2 1 0 0.4 10 Pmax , Plim [kW] m, Sω −2 − M [kgm2 ] 6 5 x [m] 8 x 10 0.6 0.8 ω [rad/s] 1 1.2 3000 2000 1000 0 0.4 0.6 0.8 ω [rad/s] 1 1.2 Added inertia, Sb ω −2 − M , and absorbed power 5 5 z [m] 0 −5 −10 −5 0 15 4000 10 5 0 0.4 10 Pmax , Plim [kW] m, Sω −2 − M [kgm2 ] 15 5 x [m] 7 x 10 0.6 0.8 ω [rad/s] 1 1.2 3000 2000 1000 0 0.4 0.6 0.8 ω [rad/s] 1 1.2 Application 2: Cylinders with simple cross sections l l l p p h p h b h O O d O e e e1 l p h c a h θ a X c a2 X p p h h O O p c O X a3 a1 a4 O e Objective functions to be minimised f1 (~x) = f2 (~x) = Z ωmax ω Z ωmin max As /Pmax (ω)dω FR max (ω)/Pmax (ω)dω, ωmin where As submerged surface area Pmax constrained maximum achievable power FR max maximum dynamic reaction force at the hinge Final Pareto fronts, ωmin = 0.4 rad/s, ωmax = 1.3 rad/s submerged elliptical cyl. surface-piercing circular cyl. submerged circular cyl. vertical flap 3 2.5 2 R FR max /Pmax dω 3.5 1.5 1 0 1 R2 3 As /Pmax dω 4 5 −3 x 10 Optimum geometries, ωmin = 0.7 rad/s, ωmax = 1.3 rad/s 1.4 a1 a4 FR max /Pmax dω 1.2 1.1 1 0.9 0.8 O 0.7 0 0.5 1 R 1.5 2 0 0 −5 −5 −10 −15 −10 2.5 3 As /Pmax dω z [m] p a3 R h 1.3 c X z [m] a2 −3 x 10 −10 −5 0 x [m] 5 10 −15 −10 −5 0 x [m] 5 10 Outline Introduction Modelling Geometry optimisation Geometry control Conclusion Recall: The inherent challenges facing a WEC I A WEC must work with large forces and low velocities I A WEC must deal with the stochastic nature of the waves What is required of a WEC to be economic? Changing wave periods A WEC must have a broad natural resonance bandwidth or otherwise be able to artificially broaden its bandwidth (by adapting itself from time to time to keep resonating with the waves). Changing wave heights A WEC must be able to avoid loads higher than the design limit. Geometry control: Examples I Changing the orientation of a WEC relative to the incident wave direction I Changing the inertia or inertia distribution of a WEC, e.g. by ballasting I Changing the freeboard of an overtopping WEC Wind turbines: Blade pitch control Maintains stable power output and mitigates loads Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Multi-objective optimisation, Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Multi-objective optimisation, adaptive geometry Can we combine these two features into one device? 1.4 1.2 1.1 1 0.9 R FR max /Pmax dω 1.3 0.8 0.5 1 R 1.5 2 0 0 −5 −5 −10 −15 −10 2.5 As /Pmax dω z [m] z [m] 0.7 0 3 −3 x 10 −10 −5 0 x [m] 5 10 −15 −10 −5 0 x [m] 5 10 Proposed WEC 2m 1.5 m X 7m 20 m O Two flap angles are considered θ = 0◦ θ = 90◦ Changing the flap angle changes its hydrodynamic properties 7 1 0 −1 0.5 30 20 1 10 1.5 ω [rad/s] 2 m55 [kgm2 ] m55 [kgm2 ] 9 x 10 2 x 10 10 0 0.5 1 10 1.5 9 d [m] 0 2 ω [rad/s] 7 x 10 x 10 30 1 20 1 10 1.5 ω [rad/s] θ = 0◦ 2 0 d [m] R55 [Nms] R55 [Nms] 20 d [m] 0 2 0 0.5 30 5 5 30 0 0.5 20 1 10 1.5 ω [rad/s] θ = 90◦ 2 0 d [m] Varying the flap angle results in differing resonant characteristics C(ω) = −m55 (ω) − Ms + 8 5 Ss ω2 8 x 10 4 x 10 C [kgm2 ] C [kgm2 ] 3 0 2 1 0 −5 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 −1 0.4 0.6 0.8 1 1.2 1.4 ω [rad/s] ω [rad/s] θ = 0◦ θ = 90◦ 1.6 1.8 2 Computed annual mean power output for different cases θ = 0◦ (vertical flap) θ = 90◦ (horizontal flap) Pann [kW] Case Description A θ = 0◦ for all sea states 46.9 B θ = 90◦ for all sea states 55.6 C Best configuration for each sea state 57.2 D θ = 0◦ for all sea states, with additional restoring force for sea states 4 to 7 62.2 Maximum reaction force for cases C (solid) and D (dashed) F R max [kN] 2000 1500 1000 500 0 1 2 3 4 5 6 7 8 Sea state C Best configuration for each sea state 57.2 kW D θ = 0◦ for all sea states, with additional restoring force for sea states 4 to 7 62.2 kW Outline Introduction Modelling Geometry optimisation Geometry control Conclusion Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Multi-objective optimisation, Research questions Modelling How to develop more realistic wave energy converter (WEC) models while at the same time reduce their simulation time? Bond graph modelling, alternative modelling of the hydrodynamic radiation term Optimisation How to incorporate the cost factor into the design problem and thus design an economical WEC? Multi-objective optimisation, adaptive geometry Further research directions I Bond graph modelling of overtopping and flexible WECs I Multi-objective optimisation of OWCs and other oscillating bodies I Multi-objective optimisation of WEC arrays I WECs with adaptive geometries Thank you