Professional Documents
Culture Documents
42 LECTURE 13:
FLUID FORCES ON BODIES
SPRING 2003
c
A. H. TECHET & M.S. TRIANTAFYLLOU
1. Morrison’s Equation
Using a two dimensional cylinder within a two-dimensional flow we can demonstrate some of the principles
we will use to determine forces on bodies in unsteady flow. Fluid particles in contact with a body surface
are motionless relative to the body so that a boundary layer is formed. From basic fluid mechanics we know
that for an inviscid flow around a cylinder we have two stagnation points at which the pressure is maximum.
In the case of a viscous flow the fluid particles follow the sides of the cylinder but as they pass a point on the
cylinder, where the pressure gradient becomes adverse, the particles will begin to separate. This separating
creates the wake which is constrained within the boundary (shear) layer originating from the separation
point.
In a turbulent flow the fluid in the boundary layer has stronger mixing, allowing the vorticity to spread
out and diffuse and for the flow to remain attached. Thus the separation point is further aft and the wake is
narrower and pressure (form) drag is reduced from the laminar case. The condition of the cylinder surface
(roughness) may also cause the early transition to turbulent flow, however in this case the frictional drag
Following the above discussion we can think of the drag as influenced by the width of the wake (narrower
wake implies less profile drag for the same diameter object) and can better understand the discussion from
lecture 12 using the system energy to determine the drag force on a body.
From last lecture the resulting force on the body takes the form of Morrison’s Equation. This is a
1
(1.1) Fx (t) = D(t) = ρCM ∇ U̇ + ρCD A U |U |
2
In order to obtain rough estimates it is advantageous to use Morrison’s equation with constant coefficients.
Suppose we want to find the estimates of the wave forces on a fixed structure. The procedure is as follows:
• 1.) Select an appropriate wave theory (linearized waves, or other higher order if necessary).
• 2.) Select the appropriate CM and CD based on Reynolds number, and other factors.
We can see from the above table that for linear waves the recommended values for drag and mass coeffi-
cients are 1.0-1.4 and 2.0 respectively. The range of drag coefficients allows us to account for roughness and
Reynolds number effects. These values are for rough estimates. In reality these coefficients vary widely with
the carious flow parameters and with time. Bretschneider showed that the values of CD and CM can even
vary over one wave cycle. Even if we ignore the time dependence of these coefficients we must account for
• Reynolds number and roughness: For smooth cylinders at Reynolds numbers around 105 laminar
flow transitions to turbulent flow, and there is a dip in CD as a function of Re. For larger Reynolds
numbers the separation point remains essentially constant and thus so does the drag coefficient. In
Roughness causes the change from laminar to turbulent flow at a lower Reynolds number and
The mass coefficient is influenced by the changes in the boundary layer and is thus also effected
u(z,t)
0
x
Suppose a vertical cylinder is subject to a current with a horizontal velocity changing both in time and
vertically in the z-direction: u(z, t). The approach in practice is to evaluate using Morrison’s formula the
force per unit length (eq. 1.2) at each point along the cylinder length and then integrate to get the total
π 2 1
(1.2) F (z, t) = CM ρ d u̇(z, t) + CD ρ d u(z, t) |u(z, t)|
4 2
Z l
(1.3) F (t) = F (z, t) dz
0
4 c
SPRING 2003
A. H. TECHET & M.S. TRIANTAFYLLOU
Z l
(1.4) M (t) = z F (z, t) dz
0
There are several limitations to these integrals. First we are limited to assume that each section does not
influence the adjacent sections’ flow. This assumption becomes questionable in the case of a cross flow that
The second assumption is that the cylinder is not piercing the free surface because in this case the water
splashing must be taken into account. In the absence of experimental data we use Morrison’s equation as a
first estimate, but we should attempt a correction for the surface piercing phenomema.
Suppose that a cylinder of diameter, d, and large length, l, is at an angle within an unsteady inflow, u(t),
and we would like to use Morrison’s equation. It has been suggested that in such cases of slender objects
First decompose the inflow velocity into two components Un and Ut where Un is the velocity normal to
the cylinder and Ut is the component tangential to the cylinder. The we can use the following expressions:
π 2 1
(2.1) Fn = CM ρ d Un + CD ρ d Un |Un |
4 2
1
(2.2) Ft = Cf ρ πd Ut |Ut |
2
to determine the normal and tangential forces per unit length. The mass and drag coefficients are found using
the diameter of the cylinder and the normal velocity as in the general (non-inclined) case. The frictional
coefficient, Cf , is used in the tangential case instead of the drag coefficient since the drag results from the
In the three-dimensional case where we have a vertical cylidner subject to a velocity vector (u, v, w) then
we must decompose this velocity into components normal and tangential to the cylinder. The convention
is to take the z-axis parallel with the cylinder axis (centerline) so that the w component is the tangential
√
component of velocity: Un = u2 + v 2 and Ut = w. Then the above equations (2.1 and 2.2) can be used.
These equations have limitations. Experiments have shown that these expressions are valid up to incline
3. Morrison’s Equation When both the body and fluid are moving
Assume that a vertical cylinder is moving with velocity u(t) within a fluid with velocity v(t), both velocities
in the horizonal direction and uniform in space, then we can write Morrison’s equation as follows:
1 π 2 π 2
(3.1) F (t) = ρ CD d h [v(t) − u(t)] |v(t) − u(t)| + ρ CM d h v̇(t) − ρ (CM − 1) d h u̇(t)
2 4 4
where d is the cylinder diameter and h the cylinder height. It is good to note that this equation does not
account for the inertial force due to the mass of the cylinder as required by Newton’s law. For example if
the cylinder was left in the free stream the we would set F (t) = m u̇(t), where F (t) is from equation 3.1 and
4.1. Integrated Force. If we consider the force caused on a vertical cylinder in depth H by a linear wave
Z 0
π 2 h 2 cosh[k(z + H)]
(4.1) FI = CM d ρ ω cos(ωt) dz
−H 4 2 sinh(kH)
h
(4.2) η(x, t) = cos(kx − ωt)
2
6 c
SPRING 2003
A. H. TECHET & M.S. TRIANTAFYLLOU
π h
(4.3) FI = ρ d2 ω 2 cos(ωt)
4 2k
Z 0
1 h2 cosh2 [k(z + H)]
(4.4) FD (t) = ρ CD d ω 2 sin(ωt) |sin(ωt)| dz
−H 2 4 sinh2 (kH)
1 h2 sinh(2Kh) + 2kH
(4.5) = ρCD d ω 2 sin(ωt) |sin(ωt)|
2 4 4k sinh2 (kH)
π h
(4.6) FIM = ρ d2 ω 2
4 2k
1 h2 sinh(2Kh) + 2kH
(4.7) FDM = ρ CD d ω 2
2 4 4k sinh2 (kH)
To compare the effects of inertial versus drag components we can take the ratio FI to FD .
FD CD h sinh(2kH) + 2kH
(4.8) =
FI πCm d 4 sinh2 (kH)
Let’s take the typical values for drag and mass coefficients, 1.0 and 2.0 respectively. We can thus find how
h 4 sinh2 (kH)
FDM > FIM when d > 2π sinh(2kH)+2kH
In the regimes where drag is unimportant (below 10% of the total force) we use the theory of inertia
dominated flows neglecting the viscous terms. When a large object is moving with stead speed near the
water free surface then it is subject to a steady force that is caused by skin friction, separation, and wave
making. The last part can be treated as an inviscid flow (wave making resistance).
FLUID FORCES ON BODIES 7
For this section let’s concentrate on the problem of forces and motions caused by waves on large objects.
We will denote the incident wave potential by φI eiωt . These are the waves with frequency ω as if no structure
existed. We denote the total potential by φ which satisfies the following relations:
(4) At the free surface, the kinematic and dynamic boundary conditions must be satisfied. These equa-
tions are nonlinear and the general solution to the resulting problem is in general impossible. For
φtt + g φz = 0 at z = 0.
(5) At infinity we must use an appropriate radiation condition. Such that the potential φ − φI must
contain only outgoing waves. Otherwise we are left with a meaningless solution of waves generated
at infinity.
As a result of these assumptions the problem becomes linear. It is to be expected, however, that the error
will be huge for cases of extreme waves and structure motion and we must correct for this when designing.
In general though for irregular ocean waves the present approach provides useful results.
Since the problem is linear we can then write the total potential as a sum of the incident, diffraction, and
radiation potentials.
The diffraction potential, φD , is caused by the presence of the structure when it is not moving, and the
The body, unconstrained in any fashion, is assumed to be free to move in six-degrees of freedom: surge,
sway, heave, roll, pitch, and yaw. These motions are denoted by xj where j = 1, 2, 3, 4, 5, 6. It is customary
8 c
SPRING 2003
A. H. TECHET & M.S. TRIANTAFYLLOU
X
6
(5.2) φR = ẋj φj
j=1
where ẋj is the velocity in the jth direction and φj is the potential caused by a unit motion in the jth
direction.
This potential decomposition is the fundamental step in vessel motion prediction and can be summarized
as follows: The total potential, φ, describing the fluid flow around a structure in the presence of waves,
equals the sum of the undisturbed wave potential, φI , the potential due to the presence of the body when it is
This will be the fundamental building block for our analysis of ship motions in waves.