1.1 The Laplace Domain
The Laplace equation is derived from the equation of continuity provided that (1) the fluid is incompressible and (2) inviscid, and (3) the flow is irrotational. Incompressibility refers to the property of the fluid to maintain its density constant within an infinitesimal volume that moves with flow velocity. An inviscid fluid, on the other hand, is a fluid with zero viscosity, namely the term (property) of the fluid that quantifies the resistance of the fluid to its gradual deformation by shear stresses. Finally, irrotational flow assumes that individual parcels of infinitesimal volume of an incompressible fluid cannot be caused to rotate. In other words, the fluid is assumed to be frictionless with no shear stresses applied on the mutual surfaces of adjacent parcels of the fluid.
Properties (1)–(3) are precisely the assumptions taken to formulate large-scale flows of liquid, namely water flows in vast volumes, just like in the open ocean, or in other words, in a marine environment. The Laplace domain has numerous applications in various disciplines of modern science, such as electromagnetics, acoustics, optics, electrostatics, imaging, hydrodynamics, and others. In hydrodynamics, however, the situation is complicated to a major extent by the existence of the free surface and subsequently by the associated boundary conditions that must be satisfied. In fact, the free surface, namely the boundary surface where the water meets the air, is what makes hydrodynamics difficult and challenging, but at the same time, a fascinating discipline.
The Laplace equation in hydrodynamics is associated with the so-called velocity potential, a scalar function, the gradient of which provides the velocity field as well as the pressure at any point of the volume of reference. Having defined the velocity field by the gradient of the velocity potential, the Laplace equation immediately follows from the mass conservation law (or the equation of continuity). Knowing the velocity potential, one may derive all dynamic parameters associated with the concerned hydrodynamic problem, namely, velocities, pressures, and accordingly hydrodynamic loads. All these, however, require the solution of relevant boundary value problems in which the boundary conditions on the free surface are of paramount importance for free-surface flows.
Describing the flow of a fluid through the Laplace equation is admittedly a simplification that leads inevitably to the assumption of an ideal fluid. This simplification, however, is a valid approach for large-scale flows in a marine environment, as has been proven beyond any doubt over the past decades. The reason is simple: in the open ocean the fluid (liquid) is water, and water is practically incompressible, nearly inviscid, while with a large degree of accuracy the flow is irrotational. In addition, use of potential theory (even nonlinear potential theory) allows the employment of analytical approaches that accordingly enable the derivation of robust analytical (or semianalytical) closed-form solutions. That advantage is provided by the separable solutions of the Laplace equation using the method of the separation of variables in various coordinate systems (Moon and Spencer, Reference Moon and Spencer1971), yielding expressions in the form of eigenfunction expansions.
It should be acknowledged, however, that all Newtonian fluids, such as water, are literally viscous. Therefore, for the completeness of the presentation, we derive in the following sections the more general formulations of the flow for Newtonian fluids, namely the Euler and the Navier–Stokes equations and eventually we focus on the flow of the ideal fluid. The analysis starts with the transport theorem, which is discussed in the section immediately following.
1.2 The Transport Theorem
We consider the following volume integral:
where is an arbitrary differentiable scalar function of position and time . The integral is taken over the given volume , which changes with time. The surface that contains is also a function of time and we assume that it has a normal velocity equal to . Let us further take the same volume integral at the time instant The change in integral is thus written as
Also, we expand in a Taylor series in to obtain
In the following, the function is denoted simply by . Retaining terms up to , (1.2) becomes
In order to simplify further the change of the volume integral , we assume that is composed of the volume and an infinitesimal volume that is contained between the adjacent surfaces and , i.e., . Hence, (1.4) is expanded in the following manner:
and it is understood that the terms also contain terms of .
To evaluate the integral over we need to consider the infinitesimal distance between and that is equal to the thickness of the small region . This thickness is the normal component of the distance traveled by in time , and is equal to the product . Hence, it is immediately deduced that the first integral in (1.5) is proportional to . In addition, given the infinitesimal thickness of volume , we may assume that the integrand is constant across that thin region in the direction normal to . Integrating in this direction only, we obtain
Taking the limit of as yields the differential form of the transport theorem, namely
The surface integral in (1.7) represents the transport of the quantity out of the volume as a result of the motion of the boundary.
1.3 Shear Stresses in Fluid Particles: The Eulerian Approach
Let us now assume a flow field in which the velocity of the fluid particles is defined by the velocity vector , where , , and (or , respectively) are the velocity components in the three directions of the Cartesian frame of reference . The density of the fluid is assumed to be constant and equal to , while the fluid particles are subjected to the external force vector that is composed of the gravitational force and the surface stresses, or force per unit area, which act on adjacent surfaces of the fluid (Figure 1.1).
We note that each surface stress should be defined by both its direction and the orientation of the surface of the fluid particle on which it is acting. For the cubical parcel of Figure 1.1 in particular, a total of stress components must be defined. The origin of the Cartesian system was chosen to coincide with one of the corners of the cube only for display purposes. The reasoning behind the tensor notation is that acts in the direction and is applied on the surface of constant , with Clearly, , denote actual shear stresses while are normal stresses (often denoted by ).
The definition of the stress components based on the cubical configuration of the fluid particles is in fact too restrictive. A more general configuration arises by assuming an infinitesimal volume of fluid (a fluid particle) in the shape of a tetrahedron with three orthogonal faces normal to the Cartesian coordinates as shown in Figure 1.2. The fourth face of the tetrahedron has an arbitrary oblique orientation. It is assumed the tetrahedron is sufficiently small and that the stresses are effectively constant along each face, and since the volume will be negligible compared with the surface area, the surface forces will dominate the body forces. To achieve an equilibrium state, the forces exerted on the four faces of the tetrahedron, which are induced by the surface stresses, must balance. By analogy with the stress theory in structural mechanics, the equilibrium state is secured by Cauchy’s law according to which there exists a Cauchy stress tensor that maps the normal to a surface to the traction vector acting on that surface. In particular, if denotes the unit normal vector on the oblique face (see Figure 1.2), Cauchy’s law implies that
or, in Cartesian representation,
The stress tensor , which is also denoted by , is written as
Accordingly, the stress component in the th direction, on a surface element with unit normal , is given by , which for convenience will be denoted by the repeating indices product
The stress tensor is symmetric, i.e., . To demonstrate that statement, we assume that the cube shown in Figure 1.1 is infinitesimal, which allows assuming that the surface moments dominate the body moments. In addition, changes in the magnitude of the stresses may be neglected across the cube and a positive value of (Figure 1.1) will cause a counterclockwise moment about the centroid. The moment on the opposite face will have the same sign, since the normal vector has the opposite sense and hence a positive acts downward. Accordingly, the counterclockwise moment is balanced by a clockwise moment that is developed only by the stress that acts in the -direction on the top and the bottom faces of the cube. Hence, it follows that and accordingly and .
1.4 Mass Conservation and Momentum Conservation
Let us now assume that , defined originally in Section 1.2, is a material volume of fluid that is assembled by a group of material fluid particles. Accordingly, the total mass of the fluid in is given by the volume integral of density in , while the mass conservation law requires that the total mass must be constant and independent of time . Hence
Further, the momentum conservation law is written as
which implies that the sum of all forces exerted on a volume of fluid must be equal to the rate of change of its momentum with respect to the Newtonian frame of reference. The momentum rate of change on the left-hand side of (1.12) refers to the th component of the velocity, the surface integral on the right-hand side denotes the sum of the stress forces in the th direction, while the last term is the total external force exerted on the concerned volume of fluid, again in the th direction. Clearly, is the th component of the aforementioned vector .
The momentum conservation law can be effectively transformed in order to be expressed by volume integrals only. This can be achieved using the Gauss theorem (also known as the divergence theorem). In particular, for a vector that is continuous and differentiable in the volume and the unit normal that is fixed on the surface surrounding and pointing to the exterior of , the Gauss theorem reads in vectorial and tensorial forms:
Hence, using (1.13) the momentum conservation law (1.12) becomes
1.5 The Equation of Continuity
Let us now consider the special case in which is a material volume always composed of the same fluid particles, while the surface in which the volume is contained moves with the same normal velocity as the fluid and . Accordingly, using (1.1), (1.7), and the Gauss theorem (1.13) it follows that
Returning to the mass conservation (1.11), (1.15) for yields
Clearly, (1.16) should hold for an arbitrary group of fluid particles and for any instant of time . Therefore the integrant itself should be zero, resulting in
Hence for an incompressible fluid and constant density, (1.17) yields readily the so-called equation of continuity, which reads
1.6 Euler Equations
If the transport theorem (1.7) is applied to the momentum conservation law (1.12), namely using instead of , it follows that
or
Again, for an arbitrary group of fluid particles and for any instant of time, (1.20) is satisfied if the integrand is zero. Thus
Finally, for a constant density and using the equation of continuity (1.18), (1.21) yields the Euler equations, i.e.,
1.7 Stress Relations in a Newtonian Fluid
Inevitably, the stress tensor and the kinematical properties of the fluid are tightly correlated. Analogous correlations hold for the stress–strain relations in solid mechanics. Assuming that the fluid is at rest, then no shear stresses are developed between the fluid particles and only a normal pressure component (pressure stress) will exist within the fluid. With reference to Figure 1.2, it is deduced that in order to balance the forces acting across a small tetrahedron, the normal pressure must be isotropic. Given that no viscous shear was considered, the stress tensor will be composed only of the pressure stress, i.e.,
By hypothesis, there are no viscous forces if the fluid moves as a rigid mass without deformation, or with the velocity field
where and are constant vectors equal to the translational and rotational velocities and is the position vector from the origin of the rotation. Viscous stresses will occur when the fluid velocity differs from (1.24) with relative motion between adjacent fluid particles. The simplest example is a uniform shear flow. Here, and for more general velocity fields, the fundamental assumption of a Newtonian fluid is that the stress tensor is a linear function of the nine gradients . This ensures vanishing of the viscous stress components for uniform translation of the fluid. The rotational term in (1.24) will be stress-free provided the gradients occur only in the form of sums . For an isotropic fluid, the values of the stress components must be independent of the choice of coordinates, and for flow in one plane there can be no shear stress in the direction normal to this plane. The most general linear function of the velocity gradients, consistent with these conditions and the requirement that is symmetric, is of the form
where is the viscous shear coefficient, also known as the coefficient of viscosity.
In order for be a tensor, the only possible addition to (1.25) for is a second constant times the divergence , which vanishes for an incompressible fluid. Thus, the total stress tensor for an incompressible fluid is given by
or in Cartesian representation
The diagonal of the first matrix in (1.27) is occupied by the normal pressure stress. The second matrix is the viscous stress tensor and is proportional to the viscous shear coefficient . The diagonal elements of the second matrix are associated with the elongations of fluid elements. The off-diagonal elements are associated with shear deformations.
1.8 The Navier–Stokes Equations
Having derived the Euler equations and expressed the stress–strain relations via (1.25) the Navier–Stokes equations are obtained by direct substitution of (1.25) into (1.22). The Navier–Stokes equations express the conservation of momentum of a Newtonian fluid. Given that the equation of continuity requires that , the derivatives of the stress tensor will read
Hence, the Navier–Stokes equations are written as
where is the coefficient of the kinematic viscosity. The Navier–Stokes equations are often written in a vectorial form according to
Finally, the Cartesian representation of the Navier–Stokes equations is
The motion of a viscous fluid with constant density and a Newtonian stress–strain relation is fully described by the system of partial differential equations (1.31)–(1.33) and the equation of continuity (1.18). Most common fluids, including water and air, appear to comply with the requirement of a Newtonian stress–strain relation. Exceptions occur for nonisotropic fluids, in which case the stress–strain relation is said to be “non-Newtonian.” Even in the former case, however, the solution of the Navier–Stokes equations raises major difficulties. Major challenges originate from the fact that those equations literally form a coupled nonlinear system that has been solved analytically only for some very simple geometrical configurations, principally those in which the nonlinear convective acceleration terms can be assumed to vanish (Newman, Reference Newman1977).
1.9 Inviscid, Incompressible Fluid and Irrotational Flow: The Velocity Potential
1.9.1 Inviscid, Incompressible
Those assumptions lead to the concept of an ideal fluid and eventually to potential flow. Incompressibility is secured by the constant density of the fluid . In addition, for a nonviscous fluid, the Navier–Stokes equations are reduced to
Further, we assume that the external force is represented by only the gravitational force ( is the gravitational acceleration), which is directed vertically downward along the coordinate. Letting the coordinate pointing upward, it follows that . Hence (1.34) may be rewritten as
1.9.2 Irrotational Flow: The Velocity Potential
The correlation of the velocity potential with the flow field is realized through its gradient, which provides the velocity vector of the fluid particles. The gain from using the velocity potential is that instead of three components (namely the velocities in all directions of the three-dimensional space) only one is needed, which apparently depends on the three spatial coordinates and the time. Further, although the velocities have an obvious physical meaning, the potential is somehow a vague term that is materialized physically through its correlation with the velocities. Nevertheless, it can be proved that the potential is indeed an effective representation of the velocity field. To this end, only the assumption of the irrotational flow is required.
In particular let us consider the definite integral
where the lower limit is some arbitrary constant position and the upper limit is the point . This integral is independent of the particular path of integration between the points and , since the difference in value of any two integrals, between the same two points, is equal to circulation around a closed path from to , along one path and back to along the other path, which is equal to zero if the fluid motion is irrotational. Thus, the integration in (1.36) can be performed along any desired path. If we choose a path that approaches the point , along a straight line parallel to the -axis, then along the final portion of the path of integration , so that
The remaining portion of the integral being a constant, does not contribute to the derivative. Applying a similar argument for the other two coordinates, we have in general that
The existence of the velocity potential therefore is univocally justified by the assumption of an irrotational flow.
Substituting (1.38) into the equation of continuity (1.18) immediately yields
Equation (1.39) is the Laplace equation given in tensorial and Cartesian representations and is the Laplace operator.
1.9.3 The Unsteady Bernoulli Equation
The Laplace equation is the governing equation of the flow field under the basic assumptions made in this section. It will also be referred to as the field equation, which we recall is the outcome of the equation of continuity. To complete the problem’s setup within the realm of potential theory we must exploit (1.35), which eventually provides the pressure field that can be also univocally obtained by the potential function only. We are particularly interested for the unsteady case, including the time derivative term of (1.35). The steady case is a special case in which this term is absent. Using (1.38), (1.35) yields
It can be easily shown that
Substituting (1.41) into (1.40) and rearranging terms yields
Finally, integrating with respect to for one gets
The integration with respect to , will produce an integration factor that is a function of time . This can be incorporated into the velocity potential, writing
Finally, using (1.44) into (1.43) yields the celebrated unsteady Bernoulli equation
The prime will be dropped in the sequel. In the Cartesian frame of reference, (1.45) is written as
Hence, the flow field for incompressible and inviscid fluid and irrotational flow is fully described by potential flow theory via the Laplace equation (1.39) and the unsteady Bernoulli equation (1.46). Those equations describe explicitly the flow in a boundless medium. Practical hydrodynamic problems, however, involve always boundary conditions that are represented by mathematical constraints, i.e., differential or algebraic equations that must be satisfied together with the Laplace and the Bernoulli equations. Even the conditions at infinity can be regarded as boundary conditions. Thus, in hydrodynamics we are invited to consider boundary value problems formed by several equations accompanying the fundamental equations (1.39) and (1.46).
1.10 Free-Surface Flow in the Laplace Domain
The hydrodynamical applications associated with the marine environment are in most of the cases fluid–structure interaction problems. In relevant situations, the factor that makes the difference is the existence of the free surface, i.e., the mutual surface that separates water and air. The effect of the free surface is of paramount importance in wave–structure interaction problems and also in hydrodynamical problems without incident waves. As an example, which will be analyzed extensively in Chapter 5, we refer to the case of a solid moving under an undisturbed free surface with no waves, namely without periodic or nonperiodic deformation of the free surface. This is the so-called “wave resistance” problem. Clearly, the motion of the solid affects the kinematical condition of the liquid up to the free surface especially in cases where the solid is in close proximity with the free surface. It is evident therefore that the free surface will affect the flow field, i.e., the velocities of the liquid particles and the pressure field exerted on the solid.
There are two main challenges associated with the existence of the free surface: (1) it imposes constraints, or in other words boundary conditions, which are in fact nonlinear and (2) the boundary where those conditions should be satisfied is time dependent and is one of the unknowns to be determined. Both issues are discussed in the present section.
In marine hydrodynamics the fluid is liquid. The domain of interest is the material volume of liquid below the free surface and the alterations of its kinematical condition(s) due to the kinematics of both the liquid (e.g., water waves, current) and the body(ies) that exist within the liquid. We start the analysis with the most typical hydrodynamical problem of a body with arbitrary geometry that floats on the free surface of a liquid field that extends to infinity both in and directions as shown in Figure 1.3 (in several cases it is most convenient to assume that the surface at infinity is a cylindrical one, situated at , where denotes the radial coordinate of a polar frame of reference). The free surface itself is a function of time and apparently of the horizontal plane coordinates . Therefore a fixed and explicitly defined plane of reference is required to define the free-surface elevation as a function of . Given the fact that in the general case the topography of the bottom, which is also a fixed surface, can be arbitrary, the only alternative is to assume as a plane of reference the undisturbed free surface and define it on . It is evident that the schematic in Figure 1.3 assumes a plane horizontal bottom that is located at a distance below the undisturbed free surface. The depth can be finite or infinite.
The control volume in Figure 1.3 is bounded from above by the free surface, coined , represented by the function , from below by the bottom surface, coined , and by the impermeable surface of the body . The basic condition is to assume infinite extend for , while one can add more complications, e.g., by restricting the extent of the control volume, which is achieved by introducing additional boundary conditions, or putting more solids into the liquid field, floating, fully immersed or bottom seated.
According to the Laplace domain approach, namely inviscid, incompressible liquid and irrotational flow, the velocity field, described by the velocity components of the liquid particles , in the three directions respectively, is expressed in terms of the scalar velocity potential as
where are the unit vectors in .
The basic hydrodynamic boundary value problem is thus defined by the following set of equations:
Equation (1.48) is the Laplace equation, (1.49) is the boundary condition that must be satisfied on the surface of the solid, (1.50) is the kinematic boundary condition on the bottom, (1.51) denotes the combined dynamic and kinematic boundary condition of the free surface, and finally (1.52) provides the exact position of the free surface. In (1.50) is the position vector of a point on measured from the center of rotation, while and are the translational and rotational vectors of the solid in surge, sway, heave, roll, pitch, and yaw respectively. Clearly represents the unit normal vector upon the associated impermeable surface that is pointing out of the fluid domain.
More detailed discussion is required for the free-surface condition (1.51). The above boundary value problem should be completed by an appropriate radiation condition at infinity, i.e., as .
1.11 Free-Surface Kinematics
The conditions associated with the presence of the free surface are the so-called dynamic and kinematic boundary conditions. Omitting the effect of the surface tension, the former condition implies that the pressure on the free surface should be continuous. The pressure at any point in the liquid is obtained through the Bernoulli equation (1.46) in which will denote the pressure of the liquid, relative to the atmospheric pressure. As the density of the air is very small compared to the density of the liquid, the motion of the air can be neglected, implying no dynamic fluctuations of the pressure on the free surface. Therefore, can be taken equal to zero and accordingly the dynamic condition on the free surface, evaluated on , becomes
Also, continuity of the interface (the surface between the air and the liquid) requires that the liquid particles on the interface remain on it during the motion of the free surface. That is expressed mathematically by
where denotes the material derivative. Next, we take the gradient and the time derivative of (1.53) to yield
and
Substituting from (1.55) into (1.54) and the resulting expression into (1.56) establishes (1.51). The exact position of the free surface is obtained directly from (1.53), which yields (1.52). Equation (1.51) is the fundamental, combined dynamic and kinematic free surface boundary condition. The fact that it has been expressed solely in terms of the velocity potential allows some simplification of the intended hydrodynamic problems, which that way, could be considered only in terms of the potential.
1.12 The Taylor Expansion of the Free Surface
The main problem associated with the free surface is that its conditions should be valid in a time-varying boundary, namely on . To overcome this difficulty, Stokes suggested the expansion of the boundary conditions (assuming small displacements of the free surface) around a mean position that coincides with the undisturbed free surface at . Therefore, any function evaluated on is written as
Accordingly, (1.52) gives for the free surface
while the boundary condition (1.51) becomes
Both (1.58) and (1.59) have been expressed on the known boundary of the undisturbed free surface on .
1.13 Expansion in Perturbations
Although the Taylor series expansion tackles the problem of the time-varying boundary in which the free-surface conditions should apply, it is unable to simplify these conditions, which remain strongly nonlinear. Admittedly, processing of nonlinear systems is a very challenging task. Consideration of (1.59) in its complete form is feasible only in the time domain. The frequency domain approximation requires further simplifications, which in hydrodynamics (and other scientific disciplines) are employed using expansions in series of perturbations (e.g., Kevorkian and Cole, Reference Kevorkian and Cole1981). The expansion of the velocity potential (and the other hydrodynamic components) in perturbation series was first employed by Stokes and accordingly the perturbation scheme is widely refereed as Stokes perturbations corresponding to the sequence of Stokes waves. Use of expansions in perturbations requires a small parameter (the scaling factor) that is the problem’s property and preferentially known. In free-surface hydrodynamics, this role is assumed by the wave steepness , where denotes the wavenumber (being equal to ) of the regular waves propagating in a liquid field of specified water depth ( is the wave length) and denotes the linear amplitude of the regular waves. Thus, using the scaling factor , the velocity potential is expressed as an infinite sum according to
There is a direct correlation between the wavenumber and the (circular) frequency of the propagating waves expressed through the so-called “dispersion” relation, which will be discussed in the sequel. This correlation involves the water depth as well, which modifies the wave length for the same wave frequency. More details will be given in the sequel. Therefore, the actual reasoning behind the expansion (1.60) in terms of the wave steepness is that the total velocity potential involves terms (powers of ) depending on the pairs , ), , and so on.
Further, assuming that at the leading (first-) order, the potential varies periodically with circular frequency , we can write
where denotes the real part of a complex argument. The nonlinear products of the inhomogeneous term in (1.59) suggest taking the following forms for the second- and the third-order potentials (phenomena higher than third-order will not be considered):
Working the same way, the free-surface elevation is expanded in a series of perturbations according to
Clearly, the second-order (double frequency) problem involves a steady component, while the third-order problem embraces also a term that varies with the leading order frequency .
The next step is to introduce (1.61)–(1.63) into the free-surface boundary condition (1.59), and (1.64) into the free-surface elevation (1.58). Doing so and equating like powers of (up to ) will result in the following boundary conditions and free-surface elevations at the various orders.
Leading (first-) order problem
Second-order problem
where the asterisk denotes the equivalent complex conjugate (cc) of the associate component.
Third-order problem
where . Higher-order problems are governed by associated relations of the form
where and denote expressions that involve terms up to order.
Equations (1.65), (1.67), and (1.70) are boundary conditions that should hold on , while (1.66), (1.68), (1.69), (1.71), and (1.72) should be evaluated on .
We recall that all potentials, at various orders, must satisfy the Laplace equation and appropriate far-field radiation conditions. Finally, the total velocity potentials should account for the boundary conditions on the fixed boundaries of the liquid field (1.49) and (1.50).
1.14 Diffraction and Radiation Potentials at the Leading Order
In the realm of the linear theory it is common to decompose the total velocity potential at the leading order into three components: the incident wave component, the diffraction component, and finally the radiation component. These are denoted by , and , respectively. The sum is usually called the Scattered velocity potential. The suggested decomposition has the meaning that any structure that is subjected to incident propagating waves causes the scattering phenomenon, which is realized by waves diffracted and radiated to infinity.
The physical acceptation of the decomposition is justified by the sequence of events that originate from the interaction of waves with the solid that exists in the wave field. Let us assume a regular wave train that is generated far away from the body, say at , after some time interacts with it and accordingly propagates far away from to the body, say to . The interaction of the regular waves with the body causes the distortion of the wave field and rationally thinking the disturbance of the flow is more profound in the vicinity of the body. Note that the surface of the body is considered rigid and non-deformable. The details of the disturbance depend on the volume of the body, its geometry and shape. The disturbance caused by slender bodies, for example, is practically negligible.
The physical intuition is that any disturbance caused by the interaction should vanish far away from the body and accordingly the velocity field and the flow should be identical in . The interaction of the waves with the solid, which is assumed to move freely, will induce oscillatory motions in all six degrees of freedom. Those motions will produce additional disturbances that are superposed to those generated by the impact of waves on the solid. All disturbances should disappear far away from the body. Therefore, the complete phenomenon can be decomposed into three subphenomena: (1) the regular wave train, represented by the incident wave potential; (2) the waves (disturbance) generated assuming that the solid is fixed and immovable (diffracted waves); and (3) the disturbance caused only by the oscillatory motions of the rigid body (radiated waves). By the linear superposition principle, the complete phenomenon is composed by a superposition of the three discrete phenomena (1), (2), and (3). As a result, mathematical synthesis requires the linear superposition (simple summation) of the factors that describe the discrete flow fields, namely the individual velocity potentials. Hence, we may write
It has been assumed that the diffracted waves are generated assuming that the structure is fixed while the radiated waves originate due to the motions of the structure in the six degrees of freedom. Hence the body condition (1.49) is accordingly decomposed into
where
Here denotes the submerged surface of the solid below the undisturbed free surface, assumed on . Further we let
where are the surge, sway, heave, roll, pitch, and yaw components of the unit normal. For harmonic oscillations one may assume that
while the radiation potential can be decomposed in terms associated with the six degrees of freedom, i.e.,
Finally, on combining together (1.76) and (1.79)–(1.81) immediately yields the boundary conditions of the six radiation (Kirchhoff) potentials in the six modes of motion, as
1.15 The Incident Wave Potential
1.15.1 Leading Order Problem
The incident wave potential constitutes the major ambient, time-varying excitation in hydrodynamical free-surface flows. The associated velocity potential at leading order has the following properties:
2. It is regular (periodic) with circular frequency in the entire liquid domain (should be an entire function of ).
3. It satisfies the kinematic condition on the bottom.
4. It satisfies the linearized (combined kinematic and dynamic) boundary condition on the free surface (1.65).
5. When the water depth is infinite, the velocity potential should be zero for ; the plus or minus sign depends on the orientation of the -axis, which is assumed fixed on the undisturbed free surface.
At the leading order the incident wave potential for infinite water depth is given by ( is assumed pointing upward)
where is the angle of propagation with respect to the horizontal -axis while and . In the general case of finite water depth , we can derive an analytic expression if the bottom is assumed flat and horizontal in the entire control volume. The bottom boundary condition for the incident wave potential thus becomes on . The potential is expressed as
where the wavenumber is given by the celebrated dispersion relation
the derivation of which will be discussed in Chapter 2.
1.15.2 Second-Order Problem
The double-frequency incident wave component should comply with the aforementioned properties (1) and (3). Also, it should be regular with circular frequency in the entire fluid domain and finally should satisfy the free-surface boundary condition at second order (1.67). The right-hand-side term of (1.67) is constructed by terms involving products of incident and diffraction potentials at the leading order. The inhomogeneous (right-hand side) term that is considered for determining should involve products of only. This term, denoted by and expressed in polar coordinates, is given by (Mei, Reference Mei1983; Malenica and Molin, Reference Malenica and Molin1995; Malenica et al., Reference Malenica, Eatock and Huang1999)
Equation (1.86) utilizes the plane gradient convention .
This yields the following form for the double-frequency incident wave component:
1.15.3 Third-Order Problem
In a similar manner, the triple-frequency incident wave component should satisfy conditions (1) and (3) of Section 1.15.1 and should be an entire function of in the entire fluid domain with circular frequency . The free-surface boundary condition associated with the third-order problem has been given in (1.70). For the incident wave problem the inhomogeneous term of (1.70) will be constructed by products of and only. Hence assuming finite water depth and introducing (1.84) and (1.87) into the right-hand-side term of (1.70) yields the inhomogeneous term of the free-surface boundary condition for the triple frequency incident wave potential. This is (Malenica and Molin, Reference Malenica and Molin1995)
while the solution for the component is
1.16 The Far-Field Radiation Condition
Also referred as the Sommerfeld radiation condition (Sommerfeld, Reference Sommerfeld1949) is the condition that must be satisfied by the diffraction and the radiation components of the potential, , . The scattered wave field must decay at infinity, so that the flow in the far field is described only by the incident wave component. The far-field condition applies directly only to the first-order terms and is valid unconditionally for arbitrary geometries. In two dimensions, with the horizontal -axis extending to infinity is expressed by
where denotes either or and is the wavenumber, which is equal to either or depending on the water depth. In three dimensions the associated radiation condition is
The radiation conditions (1.90) and (1.91) require the waves at infinity to be progressing outwards and imposes a uniqueness that otherwise would not be present. Equation (1.91) describes circular waves of decreasing amplitude as while the factor is required by energy conservation arguments.
The radiation condition for the second-order problem (diffraction potential in particular) has long been a controversial issue (Malenica and Molin, Reference Malenica and Molin1995). In fact, there is no generally accepted asymptotic formula although there have been studies that propose explicit formulations of (1.91) for the double-frequency diffraction problem (Liu and Miao, Reference Liu and Miao1991). Instead, the existing theory focuses on the asymptotic behavior of the double-frequency diffraction component, here coined . In Molin’s original analysis (Molin, Reference Molin1979) it was tentatively shown that the second-order diffraction potential, to the leading order , consists of two components: waves “locked” to the first-order wave field (also known as the “locked” wave component, or the particular solution), and “free” waves (also known as the “free” wave component or homogeneous solution) traveling in the radial direction with wavenumber obtained by the second-order dispersion relation
Even though the asymptotic behaviors of the particular and the homogeneous solutions were not established on purely rigorous mathematical foundations, they have now been accepted as being correct (Kim and Yue, Reference Kim and Yue1989). The corresponding far-field asymptotic behaviors of the homogeneous and the particular solutions are
where
Equations (1.93) and (1.94) assume that the field equation and the condition on the flat horizontal bottom have been satisfied. Analogous expressions have also been found for the third-order diffraction component and its constituent terms (see Malenica and Molin, Reference Malenica and Molin1995). A detailed analysis regarding the decomposition of the second-order diffraction potential into “locked” and “free” wave components is provided in Chapter 3.
1.17 Hydrodynamic Loading
The hydrodynamic loading on the structure is taken by integrating the hydrodynamic pressure distribution over the wetted surface. The pressure is expressed through the Bernoulli equation given in (1.46). Exciting forces and moments require that the structure is considered to be fixed. Integration of pressure over the surface yields
where, as usual, denotes the generalized direction being considered. In order to collect the terms at the different orders of wave steepness , the integral on the wetted surface is decomposed according to
where is the mean wetted surface and the mean waterline. In (1.97), , , and denote surface integrals while denotes a line integral. Equation (1.97) has the meaning that the pressure is integrated over the mean wetted surface plus an additional part to account for the free-surface elevation above the undisturbed free surface. The expansion in perturbations (1.60) and (1.64) suggests taking the following series expansion for the hydrodynamic loading as well:
Hence the following expressions for the time varying hydrodynamic loading up to the first three orders are derived:
1.18 Added Mass and Hydrodynamic Damping Coefficients
The component of the hydrodynamic loading that arises from the pressure integration associated with the forced oscillations with frequency and the radiation component can be written as (see also Newman, Reference Newman1977, Section 6.17; Linton and McIver, Reference Linton and McIver2001, Section 1.3.4)
The factor , being by default a complex number, is convenient to be decomposed into a real and an imaginary part as
yielding
The term is known as the added mass coefficient given that it represents a force that is proportional to acceleration. In contrast, represents a force that is proportional to the velocity of the body and thus it is called the hydrodynamic damping coefficient. In fact they represent the associated hydrodynamic parameter in direction due to the oscillatory motion of the body with frequency and small amplitude in the direction .
From the physical point of view the added mass is a virtual mass that depends on the geometry and the volume of the body, and apparently the density of the liquid, and expresses the added inertia introduced into the oscillatory system because of the volume of liquid that is displaced as the body is accelerating. In other words, in order for the body to move inside the liquid must first accelerate some volume of it. In the same manner, the hydrodynamic damping coefficient expresses a quantity of added damping introduced into the system. The hydrodynamic damping is defined in the form of a linear component and is directly proportional to the velocity.
1.19 The Green’s Theorem
The Green’s theorem follows from Green’s second identity that is derived in the following manner. Let and be continuously differentiable functions in a region bounded by a surface . Both and have continuous partial derivatives of the second order in . Then the pair of functions and satisfy the following (Kellogg, Reference Kellogg1967; p. 212)
If we interchange and and the resulting equation is subtracted from (1.105) the result is Green’s second identity, namely
where is a coordinate directed in the outward normal direction to the surface . If and are both solutions of the Laplace or Helmholtz equation, the latter being given by
in the region , for an arbitrary [for , (1.107) yields the Laplace equation] then (1.106) is reduced to
which is widely referred as the Green’s theorem.
Let us next consider the two-dimensional problem of the function at a particular point , , so that the domain is some surface bounded by the closed curve , as shown in Figure 1.4. We use capital letters to denote points within and lowercase letters to denote points on the boundary of . The distance between and an arbitrary point within is denoted by , while the Green’s function is now denoted by . The Green’s function must be singular at , i.e., for .
Our goal is to calculate the value of the function at point , and therefore we consider a point upon , coined . Given that is singular as , the point must be excluded from region . Therefore, we define a small circle embracing , with radius . Further, use is made of the Green’s theorem to and over the region between and . Thus the line of integration is and we write
Here, denotes the outward normal to the surface located on the point . Elaborating further (1.109) we obtain
The second term in the left member of (1.110) results in the value if behaves in such a way that
and
and yet is finite. In particular (1.112) should give , while it is recalled that must be singular for . All these requirements are satisfied by taking
which implies a logarithmic singularity at . Finally, (1.110) yields the following integral representation for :
If is taken outside , then it holds that
Further, we discuss some properties of the integral representation (1.114). In hydrodynamics, the velocity potential typically satisfies Neumann or Dirichlet boundary conditions on a line contour that bounds a region , in the form or , respectively, for all . Note the linear free-surface boundary condition (1.65) is of Robin type. In relevant cases, (1.114) is reduced to single-term integration. Further, if is known on and can be constructed in such a way so that for all , then is obtained explicitly. Accordingly if is known on and for all , then, again, is obtained explicitly. If the point (say point ) in which we want to calculate lies on the contour line (see Figure 1.4), then to apply the Green’s theorem we must exclude that point and we must assume that it is enclosed by a semicircle instead of a circle. Therefore, application of the Green’s theorem over the domain and the semicircle in will yield the following integral representation for :
An integral equation for the boundary values of (points lying on ) arises from (1.115) if is known for all .
Equation (1.116) allows some simplifications if the Green’s function can be constructed in such a way that satisfies the same boundary conditions as on a subinterval . Provided that satisfies Neumann or Dirichlet conditions on , the contribution of the integral over vanishes.
Equation (1.116) is used to calculate on all points lying on . The next step, after solving the integral equation, is to employ (1.114) to calculate on all points inside the region .
In particular problems the Green’s function may be symmetrical, i.e., . Hence from (1.114) and (1.116) we have
Integral equations (1.117)–(1.118) may now be interpreted in terms of distributions of sources and dipoles over the boundary surface .
The same approach can be taken for the three-dimensional problem where is a volume bounded by the surface . The analysis and the equations for the three-dimensional problem are explicitly the same with those presented previously that referred to the two-dimensional space, provided that the integrals over are regarded as surface integrals. To exclude the point from surface , we should assume a sphere containing , while when the point lies on the surface (denoted by ) it should be bounded by a hemisphere. The Green’s function in the three-dimensional case will read
which is evidently singular for .