Maximum wave-power absorption under motion constraints D. V. E V A N S School of Mathematics, University of Bristol, Bristol BS8 1 714/,, UK An expression is derived for the maximum mean power that can be absorbed by a system of oscillating bodies in waves under a global constraint on their motions. The particular case of a single half-immersed sphere is used to show how the 'point absorber' result predicting capture widths in excess of unity must be modified. The theory is also applied to the submerged cylinder wave-energy device and curves are presented which show how the maximum efficiency is affected by restricting the motion of the device. INTRODUCTION Results for the maximum power which a single body in two or three dimensions can absorb when oscillating in resonance with an incident sinusoidal wave train have been given by Evans ~, Mci 2 and Newman 3. More recently these results have been generalized to an arbitrary number of bodies by Evans 4 and Falnes 5. In each case it was assumed that the body was free to oscillate with whatever amplitude was necessary to achieve the optimal conditions for maximum power. An exception to the papers mentioned is that due to Newman 6 who imposed motion constraints in considering the power absorption capability of a slender elongated body. In practice most wave-energy devices will have physical limitations placed upon their excursions due to restraints such as mooring lines or pump stroke. Whereas there exists a significant band of wave frequencies and lengths for which the device does not reach these limits, in long waves a relatively small device would, ideally, need to make large excursions to achieve optimum power absorption. Again larger amplitude waves require larger device excursions for optimum absorption. Here, simple expressions are presented for the maximum efficiency of power absorption of a body when its velocity and hence amplitude is constrained such that its magnitude never exceeds a given multiple of the incident wave amplitude and also for a system of absorbing bodies under global velocity constraints. Examples of a floating hemisphere in three dimensions and a submerged circular cylinder in two dimensions are presented. damping matrix. Let X be the complex N-vector denoting the amplitude and phase of the exciting force on each body in its direction of subsequent motion, when all bodies are held fixed. Then it can be shown 4'5 that: P{U)=¼(x*u + u*x)-½u*Bu : 1X*B- I X - ~ ( U - ½ n - 1X ) * B ( U - ½B - 1X) (1) where * denotes conjugate transpose, and the second expression follows from the first by elementary manipulation provided B 1 exists. It is clear that the maximum value of P when U is unconstrained is given by the first term in equation (11 when U =½B-~X. We shall assume that IIuII ~- u * u ~</~2 (2) and we shall optimize equation (1) subject to equation (2). It is convenient to define Y=½B 1X so that equation (1) becomes: P(U) = ½ Y ' B Y - ~ U - Y)*B(U - V ) {3) Clearly if Y*Y ~<f12 then Uop, = Y giving Pmax(U)=½Y'BY (=~X*B 1X). IfY*Y>~fl z, Uop~is given from geometrical considerations. Thus Pmax(U)is the largest value of C for which the closed surfaces P(U) = C touch the surface U*U =/32, in the complex N-dimensional U-space. This is most easily determined by introducing a Lagrange multiplier #. Consider the scalar: Q(U) = P( U ) - ½#{U*U - f12) THEORY We consider N independently oscillating bodies making simple harmonic motions of radian frequency o~ in one degree of freedom in response to a long-crested incident wave-train also of frequency ~. Each body can absorb energy from the incident wave field. Let U be the complex N-vector denoting the amplitude and phase of the velocity of each body, and B the N x N real, symmetric radiation 0141 1187/81/040200-0452.00 ©1981 C M L P u b l i c a t i o n s 200 Applied Ocean Research, 1981. Vol. 3, No. 4 A necessary condition for maximizing equation (3) subject to U*U = fi2 is that V~o(U)= 0 where V denotes the 2Ndimensional vector gradient operator involving the partial derivatives of both real and imaginary parts of Ui for each i. It follows that: -B(U- Y)-pU =0 Maximum wave-power absorption under motion constraints: D. V. Evans Thus or U = Uopt = (B + #I)- 1BY Pmax (4) Substitution of this value back into U*U=/~ 2 gives a single scalar equation to determine/a: 1 2[1-H(1-f)(1-6)2]/B =~lX,I where = Y*B(B +/al)- 2By ~- ¼X*(B + #I)- 2X =/32 ----II¥*FR( B +LV p l ) t- 2--/ B ~ 2- ] X s ~ , - s for /~ ~<½lIB- 1X, l[- IIYII and P ~ , ( U ) --~..~ 1 Y ' B - 1X~ for /~>~{IIB-1X~lI B/(B + 2~BIXI- Ia) = (10) ' (5) and H(x) is the Heaviside step function. For a vertically oscillating buoy with a vertical axis of symmetry it can be shown that: It follows from equations (4) and (3) that: Pmax(U) = ½U*(B + 2/~I)U (9) (6a) 2rrlX[ 2 = 8B2P w (6b) where Pw = ½P9A2.0)/29 is the mean power per unit crestlength of the incident wave, and 2 is the incident wave length. We can define a capture width I= P/Pw so that: (7) 2 ema,/P~, ----/max= 2~{ 1 -- H(1 - 6)(1 -- 6) z } Clearly Pma,(U) is continuous in ft. where 5 = 20)AotB/IX[ = o~. EXAMPLES In practice each of the N bodies would have constraints on their amplitudes and velocities which may be of the form: lUll ~<0)A0¢i=fli (i = 1, 2 ..... N) (8) where A is the incident wave amplitude and ~i are constants governing the permissible body amplitude compared to the incident wave amplitude. In normal operating conditions ~ = 0 ( 1 ) but in larger waves ~ may be small. It is not obvious how the N constraints (8) can be handled analytically and in particular cases a numerical optimisation procedure would have to be used. The global constraint (2) goes some way towards the solution, in the sense that with fl = min{fl~, i = 1, 2 ..... N} then if equation (2) is satisfied so is equation (8) since: IUilZ<~u*u<~3z<~3[ Pmax klXI 2 E1 p2/(B +/~)2]/B = for - /~ ~<½IB 'XI where IX[ = 2/~(B + #) determines/~. (11) where K = 2rt/2 = 0)2/0 is the wave number, a the radius of the buoy, and 2aa=B/M0) the damping coefficient in heave, non-dimensionalized with respect to a typical mass, m. Figure 1 shows the variation of lmax/2a with Ka for different c¢ for a half-immersed sphere with M=2rrpa 3 where values of).33(Ka) have been taken from Havelock 8. It can be seen that if the amplitude of motion of the sphere is not allowed to exceed the wave amplitude the capture width never exceeds 70% of the sphere diameter in [ II~ 1.4 1.2 a=2 (i=1, 2 . . . . . N) The global constraint applied here may be more appropriate to devices such as the French 7 flexible bag where a fixed air volume is driven through a turbine and re-circulated due to the compression of segmented flexible bags. To apply the above theory to more than one oscillating body requires knowledge of B and X for such configurations and accurate values are not available even for the simplest cases. Instead we shall concentrate on a single body in two or three dimensions. For a single body oscillating in one degree of freedom all vectors and matrices become scalars and, from equation (6b): (2M/pa3)U2(Ka)3/2).13/32 I 1-0 o,8 06 0'4 0-2 [ 0 0"4 15 10 I I I 0"8 6 [ 4 1 1'2 3 I Ka I I I 1"6 2"0 2"4 2 I 1-5 I 2 8 X/2a Figure 1. Variation of maximum capture width/diameter ratio with wavelength for half immersed energy-absorbin 9 sphere for different motion constraints. , unconstrained sphere Applied Ocean Research, 1981, Vol. 3, No. 4 201 Maximum ware-power absorption under motion constraints: I). li Et'an.~ where 14"[(14',.) is a wave-making coefficient being the ratio of the wave amplitude far upstream {downstream) caused by unit amplitude oscillatory motion of the body. For instance, for a symmetric body for which 14 +]-]A l- 1.o t/ ..... = ½ [ 1 where ~- 0'6 - H(1 - 5 ) ( 1 - ,~)2] (16) (17) ,5 = 2c~W,. and ~i- is the wave-making coefficient. For a symmetric section absorbing energy through oscillations in two uncoupled degrees of freedom 0,4 2 0.2 J t/max= ]--21 E i 0'2 0 0"4 i 1 0"6 0"8 1!0 1.i2 ga 3020 t i 12 8 6 4 3 I I I I i "X/2a Figure 2. Variation in loss of maximum e/[ficiency of an energy-absorbing submerged circular cylinder with wavelength for different motion constraints. Submergence c .I a contrast to the result from unconstrained motions where /max( --= '~/2Yt) increases indefinitely as Ka--,O. For sphere amplitudes up to twice the wave amplitude (~=2) /max peaks at a value 8% greater than the sphere diameter showing that energy is being absorbed from outside the region occupied by the sphere. In steeper waves where may have to be as small as 0.5 the capture width never exceeds 40% of the sphere diameter. Note that we are not concerned here with how the power is absorbed. In particular the difficulty in implementing the required spring rate for optimal absorption in this case, as discussed by Evans ~, is not considered. Many wave-energy devices begin life as cylindrical sections spanning a narrow wave-tank, so it is appropriate to consider the form of equation (9) in this case. Now for a tank of width Lwe have IXl and B =½PcolA +127 tL where 6~=2~iW~ in an obvious rotation. Now a submerged cylindrical section making a combination of vertical and horizontal oscillations has been proposed as a potentially attractive wave-energy device 9. In this case W~,I = W,,2 in deep water and t/m~,= 1 - H ( 1 - c~)(1_,5)2 (18) with ,5 as in equation (17). Here we have made the not unreasonable assumption that the constraints on vertical and horizontal motions of the cylinder are the same so that % =~2 =~. The result (18) can be derived more directly by noting that a combination of horizontal and vertical motions n/2 out of phase can give rise to a wave-train travelling downstream only, of maximum amplitude 2~AW,,. The incident wave is not reflected by the cylinder and the resulting transmitted wave due to the superposition of the incident wave with the downstream radiated wave has minimum amplitude A(I - 2~W~). Energy considerations now give equation (18). Similarly equation (14) can be derived directly starting with Evans ~ equation (3.7). It is clearer in this case, where the maximum efficiency in the unconstrained motion is 1, to graph the maximum efficiency loss At/ where At/= 1 - t/m~,-=H(I -6)(1 --0) 2 Figures 2 and 3 show the variation of At/ with 1.0 (13) where A +(A-) is the complex amplitude of the potential far upstream, in the opposite direction to the incident wave (downstream) due to the forced oscillatory motion of the cylindrical section with unit velocity. Also 7={l+IA-/A+]2} -1. In this case we can define an efficiency t/so that: where _@2 (I 2) = pgAIA + IL t/m~,-- Pm~/P,,,-/[1 - H ( 1 --b)(1 H(I-0,)(I t --5) 2] 0.8 c~ <3 0'6 0"4 (14) 0.5 0.2 5 = 2~(M/pa2L)1:2Ka(27)- 1.,2)" 31:2 3 J and equations (12) and (13) have been used in equations (9) and (10). A simpler version of equation (14) in terms of A + is: 0 0"2 0"4 0.'8 0'6 1!o , !2 Ka 3020 L I 12 I 8 I 6 I 4 I 3 I kl2a 5=~KIA+]/7 =o~w) 202 /, , Applied Ocean Research, 1981, Vol. 3, No. 4 (15) Figure 3. Variation in loss ¢?f maximum eJficiency of'an energy-absorbing circular cylinder with wavelength for d!fferent motion constraints. Submergence c = a M a x i m u m wave-power absorption under motion constraints: D. V. Evans w a v e n u m b e r for different values of~ and for two depths of submergence. In Fig. 2 the submergence of the top of the cylinder is c = a / 4 . F o r small waves, with ~ = 2 , say corresponding to a m a x i m u m cylinder motion of twice the wave amplitude a loss in m a x i m u m efficiency only becomes apparent for wavelengths in excess of about 18 times the cylinder diameter. In bigger waves, a smaller value of ~ is appropriate and then the loss in efficiency occurs at much shorter wavelengths. F o r instance with =0.25 there is a loss in efficiency at all incident wavelengths reflecting the fact that the cylinder is incapable of radiating a wave large enough to cancel the incident wave. One would anticipate a reduced efficiency for another reason, namely the tendency for bigger waves to steepen and break up into higher harmonics past the cylinder 9. Figure 3 shows similar results for a more deeply submerged cylinder with clearance c = a. As expected the loss in efficiency occurs more readily as the wave-making capability of the cylinder is further reduced. CONCLUSIONS The results confirm that three-dimensional small point absorber devices can only achieve capture widths in excess of unity if amplified b o d y motions are allowed. F o r long terminator devices the importance of the wavemaking capability of the device is illustrated by considering a long energy absorbing cylinder submerged to different depths. Linear theory has been used t h r o u g h o u t so that the constraints on the b o d y motion characterized by the n u m b e r :t must assume to be applied by an increased linear d a m p i n g force. In practice, however, such constraints would invariably be non-linear and the above theory would need modification. REFERENCES 1 Evans, D. V. A theory for wave-power absorption by oscillating bodies, J. Fluid Mech., 1976, 77, 1 2 Mei,C. C. Power extraction from water waves, J. Ship. Res., 1976,20, 63 3 Newman, J. N. The interaction of stationary vessels with regular waves, Proc. l lth Symp. Naval Hydrodynamics, London, 1976, p. 491 4 Evans, D. V, Some theoretical aspects of three-dimensional waveenergy absorbers, Proc. 1st Symp. Ocean Wave Eneroy Utilization, Gothenberg, 1979 5 Falnes, J. Radiation impedance matrix and optimum power absorption for interacting oscillators in surface waves, Appl. Ocean Res., 1980, 2, 75 6 Newman, J. N. Absorption of wave energy by elongated bodies, Appl. Ocean Res., 1979, 1, 189 7 French, M. J. Hydrodynamic basis of wave energy converters of channel form, J. Mech. Eng. Sci. (C) I. Mech. E., 1977. 19 8 Havelock, T. H. Waves due to a floating sphere making periodic heaving oscillations, Proc. R. Soc., 1955, (A), 231, 1 9 Evans, D. V., Jeffrey, D. C., Salter, S. H. and Taylor, J. R. M., Submerged cylinder wave-energy device: theory and experiment, Appl. Ocean Res., 1979, l, 3 Applied Ocean Research. 1981, Vol. 3, N o 4 203