A
n important area in momentum transfer is inviscidflow, in which, by virtue of the absence of shear stress, analytical solutions to the differential equations offluidflow are possible.The subject of inviscidflow has particular application in aerodynamics and hydrodynamics and general application toflow about bodies—the so-called external flows. In this chapter, we shall introduce the fundamentals of inviscidflow analysis.
▶ 10.1
FLUID ROTATION AT A POINT
Consider the element offluid shown in Figure 10.1a. In timeDtthe element will move in thexyplane as shown. In addition to translation, the element may also deform and rotate.
We have discussed the deformation previously in Chapter 7. Now let us focus our attention on the rotation of the element. Although the element may deform, the orientation will be given by the average rotation of the line segmentsOBandOAor by denoting the rotation by
wz d dt
ab 2
where the counterclockwise sense is positive. From Figure 10.1a, we see that
O y
x t A
B
O
A B
Figure 10.1a Rotation of afluid element.
124
wz lim
Dx;Dy;Dz;Dt®0
1 2
arctanf vyjxDx vyjxDt=Dxg Dt
arctanf vxjyDy vxjyDt=Dyg Dt
which becomes, in the limit,
wz1 2
@vy
@x
@vx
@y
(10-1) The subscript zindicates that the rotation is about thezaxis.
In the xzand yzplanes the rotation at a point is given by wy1
2
@vx
@z
@vz
@x
(10-2) and
wx1 2
@vz
@y
@vy
@z
(10-3) The rotation at a point is related to the vector cross product of the velocity. As the student may verifyÑ
Ñv @vz
@y
@vy
@z
ex @vx
@z
@vz
@x
ey @vy
@x
@vx
@y
ez
and thus
Ñv2w (10-4)
The vectorÑvis also known as thevorticity.When the rotation at a point is zero the flow is said to be irrotational.For irrotational flow Ñv0, as can be seen from equation (10-4). The significance offluid rotation at a point may be examined by a different approach. The Navier–Stokes equation for incompressibleflow, equation (9–29), may also be written in the form
rDv
Dt ÑPrg mÑ Ñv (9-29)
It may be observed from the above equation that if viscous forces act on afluid, theflow must be rotational.
Figure 10.1b illustrates the effect of vorticity, and whether thefluid element is rotational or irrotational. The vorticity is often described as the measure of the moment of momentum of a small differential area about its own center of mass. The vorticity vector is characterized mathematically as the curl of the velocity vector. In Figure 10.1b, a velocity gradient is
Figure 10.1b Rotation of afluid element as the result of a velocity gradient.
10.1 Fluid Rotation at a Point ◀ 125
moving along aflat plate. Near the plate, as a result of the no-slip boundary condition, the velocity of thefluid is zero at that wall and increases into the bulkfluid. Near the wall, a small differential area offluid is acted upon by the velocity gradient such that at the bottom of thefluid element, the velocity will be less than at the top of the element resulting in a net rotation or spin of the element in the clockwise direction as shown by the arrow in the center of the differential area. The result of this is the rotation of thefluid element and rotational flow. Out in the bulk, the velocity is the same at the top and bottom of thefluid element, resulting in the differential element having no net rotation—thus the term irrotationalflow.
Vorticity has many applications in weather patterns, such as in tropical cyclones1and in flow of blood in the body.
The kinematic condition Ñv0 is not the first time we have encountered a kinematic relation that satisfies one of the fundamental physical laws offluid mechanics.
The law of conservation of mass for an incompressibleflow,Ñ
v0, is also expressed as a kinematic relation. The use of this relation is the subject of the next section.Example 1
A two-dimensional velocity is given by the equation
v 6yex 6xey
Is theflow rotational, or irrotational?
To be irrotational, theflow must satisfy the equationr v0, and as a result, Equation 10.1 must be equal to zero.
wz1 2
@vy
@x
@vx
@y
0
First, we mustfind values for@vy/@xand@vx/@yto see if the condition of (equation (10-1)) is satisfied.
From the given equation, wefind the velocity components:
vx6y and vy6x Next we take the necessary partial derivatives dictated by equation (10-1):
@vx
@y 6
@vy
@x 6 And thus, plugging in the values,
wz1
2 6 6 0 So the condition of irrotationalflow is satisfied, and this particle will not rotate.
Let’s also examine another different, but similar, condition. A two-dimensional velocity is given by the equation v 6yex 6xey
Is thisflow rotational, or irrotational? We must again see if equation (10-1) is satisfied.
Finding values for@vy/@xand@vx/@yas we did previously, vx6y and vy 6x
1J. C. L. Chan,Ann. Rev. Fluid Mech.,37, 99 (2005).
Next we take the necessary partial derivatives dictated by equation (10-1):
@vx
@y 6
@vy
@x 6 giving
wz1
2 6 6 6≠0 So the condition is not satisfied, and this is an example of rotationalflow.
▶ 10.2
THE STREAM FUNCTION
For a two-dimensional, incompressible flow, the continuity equation is
Ñ
v@v@xx@v@yy0 (9-3)Equation (9-3) indicates that vx and vy are related in some way so that @vx/@x (@vy/@y). Perhaps the easiest way to express this relation is by having vx and vyboth related to the same function. Consider the function F(x,y), ifvxF(x,y), then
@vy
@y @F
@x or vy
Z @F
@xdy
Unfortunately, the selection of vxF(x, y) results in an integral for vy. We can easily remove the integral sign if we make the original F(x,y) equal to the derivative of some function with respect to y.For example, ifF(x,y)(@Y(x,y)/@y], then
vx@Y
@y As @vx/@x (@vy/@y), we may write
@vy
@y @
@x
@Y
@y or @
@y vy@Y
@x
0
for this to be true in general:
vy @Y
@x
Instead of having two unknowns, vx and vy, we now have only one unknown, Y. The unknown, Y, is called the stream function. The physical significance of Y can be seen from the following considerations. As YY(x,y), the total derivative is
dY@Y
@xdx@Y
@ydy
10.2 The Stream Function ◀ 127
Also
@Y
@x vy and @Y
@y vx
and thus
dY vydxvxdy (10-5)
Consider a path in thexyplane such thatYconstant. Along this path,dY0, and thus equation (10-5) becomes
dy
dxjYconstantvy
vx
(10-6) The slope of the pathYconstant is seen to be the same as the slope of a streamline as discussed in Chapter 3. The functionY(x,y) thus represents the streamlines. Figure 10.2 illustrates the streamlines and velocity components forflow about an airfoil.
The differential equation that governs Y is obtained by consideration of the fluid rotation,w, at a point. In a two-dimensionalflow,wz12 @vy=@x @vx=@y;and thus, if the velocity componentsvyandvxare expressed in terms of the stream functionY, we obtain, for an incompressible, steadyflow,
2wz@2Y
@x2 @2Y
@y2 (10-7)
When theflow is irrotational, equation (10-7) becomes Laplace’s equation:
Ñ2Y@2Y
@x2 @2Y
@y2 0 (10-8)
Example 2
The stream function for a particularflow is given by the equationY6x2 6y2. We wish to determine the velocity components for thisflow, andfind out whether theflow is rotational or irrotational.
We defined the stream function as
@Y
@x vy
Figure 10.2 Streamlines and the stream function.
and
@Y
@yvx
Thus,
vx@Y
@y 12y vy @Y
@x 12x The equations for the velocity components arevx 12yandvy 12x.
Next, we want to determine whether theflow is rotational or irrotational. To do this, we must satisfy equation (10-1).
wz1 2
@vy
@x
@vx
@y
0 Solving for the necessary partial derivatives,
@vy
@x 12
@vx
@y 12 Thus,
wz1
2 12 12 0 and thisflow is seen to be irrotational.
▶ 10.3
INVISCID, IRROTATIONAL FLOW ABOUT AN INFINITE CYLINDER In order to illustrate the use of the stream function, the inviscid, irrotational flow pattern about a cylinder of infinite length will be examined. The physical situation is illustrated in Figure 10.3. A stationary circular cylinder of radiusais situated in uniform, parallelflow in the xdirection.
As there is cylindrical symmetry, polar coordinates are employed. In polar coordi- nates,2equation (10-8) becomes
@2Y
@r2 1 r
@Y
@r1 r2
@2Y
@q2 0 (10-9)
where the velocity componentsvrandvq are given by vr 1
r
@Y
@q vq @Y
@r (10-10)
2The operatorÑ2in cylindrical coordinates is developed in Appendix A.
10.3 Inviscid, Irrotational Flow about an Infinite Cylinder ◀ 129
The solution for this case must meet four boundary conditions. These are as follows:
1. The circleramust be a streamline. As the velocity normal to a streamline is zero, vrjra0 or@Y=@qjra0:
2. From symmetry, the line q0 must also be a streamline. Hence, vqjq00 or
@Y/@r|q00.
3. Asr®
∞
the velocity must befinite.4. The magnitude of the velocity asr®
∞
isv∞, a constant.The solution to equation (10-9) for this case is
Y r;q v∞rsinq 1 a2 r2
(10-11) The velocity componentsvrand vqare obtained from equation (10-10):
vr 1 r
@Y
@q v∞cosq 1 a2 r2
(10-12) and
vq @Y
@r v∞sinq 1a2 r2
(10-13) By settingrain the above equations, the velocity at the surface of the cylinder may be determined. This results in
vr0 and
vq 2v∞sinq (10-14)
The velocity in the radial direction is, of course, zero, as the cylinder surface is a streamline.
The velocity along the surface is seen to be zero atq0 andq180°. These points of zero velocity are known asstagnation points.The forward stagnation point is atq180°, and the aft or rearward stagnation point is at q0°. The student may verify that each of the boundary conditions for this case are satisfied.
r y
a x
Figure 10.3 Cylinder in a uniformflow.
▶ 10.4
IRROTATIONAL FLOW, THE VELOCITY POTENTIAL
In a two-dimensional irrotational flow Ñv0, and thus @vx/@y@vy, /@x. The similarity of this equation to the continuity equation suggests that the type of relation used to obtain the stream function may be used again. Note, however, that the order of differentiation is reversed from the continuity equation. If we let vx@f(x,y)/@x, we observe that
@vx
@y @2f
@x@y@vy
@x or
@
@x
@f
@y vy
0 and for the general case
vy@f
@y
The functionfis called thevelocity potential. In order forfto exist, the flow must be irrotational. As the condition of irrotationality is the only condition required, the velocity potential can also exist for compressible, unsteadyflows. The velocity potential is commonly used in compressibleflow analysis. Additionally, the velocity potential,f, exists for three-dimensionalflows, whereas the stream function does not.
The velocity vector is given by
vvxexvyeyvzez@f
@xex@f
@yey@f
@zez
and thus, in vector notation,
vÑf (10-15)
The differential equation definingfis obtained from the continuity equation. Considering a steady incompressible flow, we have Ñ
v0; thus, using equation (10-15) for v, we obtainÑ
ÑfÑ2f0 (10-16)which is again Laplace’s equation; this time the dependent variable isf. Clearly,Yandf must be related. This relation may be illustrated by a consideration of isolines ofYandf. An isoline of Y is, of course, a streamline. Along the isolines
dY@Y
@xdx@Y
@ydy or
dy dx
Yconstantvy
vx
10.4 Irrotational Flow, the Velocity Potential ◀ 131
and
df@f
@xdx@f
@ydy dy dx
df0
vx
vy
Accordingly,
dy=dxjfconstant 1 dy=dx
Yconstant
(10-17) and thusYand forthogonal. The orthogonality of the stream function and the velocity potential is a useful property, particularly when graphical solutions to equations (10-8) and (10-16) are employed.
Figure 10.4 illustrates the inviscid, irrotational, steady incompressible flow about an infinite circular cylinder. Both the streamlines and constant-velocity potential lines are shown.
Example 3
The steady, incompressibleflowfield for two-dimensionalflow is given by the following velocity components:vx16y xand vy16xy. Determine the equation for the stream function and the velocity potential.
First, let’s check to make sure continuity is satisfied:
@vx
@x@vy
@y @
@x 16y x @
@y 16xy 110 (1)
So continuity is satisfied, which is a necessary condition for us to proceed.
We defined the stream function as
@Y
@yvx (2)
and
@Y
@xvy (3)
Figure 10.4 Streamlines and constant- velocity potential lines for steady, incompressible, irrotational, inviscidflow about a cylinder.
Thus,
vx@Y
@y16y x (4)
vy @Y
@x16xy (5)
We can begin by integrating equation (4) or equation (5). Either will result in the same answer. (Problem 10.24 will let you verify this.) We will choose to begin by integrating equation (4) partially with respect toy:
Y8y2 xyf1 x (6)
wheref1 xis an arbitrary function ofx.
Next, we take the other part of the definition of the stream function, equation (3), and differentiate equation (6) with respect tox:
vy @Y
@xy f2 x (7)
Here,f2 xisdf
dx, sincefis a function of the variablex.
The result is that we now have two equations forvy, equations (5) and (7). We can now equate these and solve for f2 x:
vyy f2 x 16xy Solving forf2 x,
f2 x 16x So
f1 x 16x2
2 8x2C
The integration constant C is added to the above equation sincefis a function ofxonly. Thefinal equation for the stream function is
Y8y2 xy 8x2C (6)
The constant C is generally dropped from the equation because the value of a constant in this equation is of no significance.
Thefinal equation for the stream function is
Y8y2 xy 8x2
One interesting point is that the difference in the value of one stream line in theflow to another is the volumeflow rate per unit width between the two streamlines.
Next we want tofind the equation for the velocity potential. Since a condition for the velocity potential to exist is irrotational flow, we mustfirst determine whether theflow in this example is irrotational.
To do this, we must satisfy equation (10-1).
wz1 2
@vy
@x
@vx
@y
1
2 16 16 0 So theflow is irrotational, as required.
We now want to determine the equation for the velocity potential. The velocity potential is defined by equation (10-15):
v rf
or @f
@xvx16y x
10.4 Irrotational Flow, the Velocity Potential ◀ 133
f16xy x2 2f y
Differentiating with respect toyand equating tovx,
@f
@y16x d
dyf y 16xy Thus,
d
dyf y y and
f y y2 2 so that thefinal equation for the velocity potential is
f16xy x2 2y2
2
▶ 10.5
TOTAL HEAD IN IRROTATIONAL FLOW
The condition of irrotationality has been shown to be of aid in obtaining analytical solutions influidflow. The physical meaning of irrotationalflow can be illustrated by the relation between the rotation or vorticity,Ñv, and the total head,P/rv2/2gy. For an inviscid flow, we may write
Dv
Dtg ÑP
r Euler’s equation
and
Dv Dt @v
@t Ñ v2
2 v Ñv Vector identity
As the gradient of the potential energy is g, Euler’s equation becomes, for incompressible flow,
Ñ P rv2
2 gy
v Ñv @v
@t: (10-18)
If theflow is steady, it is seen from equation (10-18) that the gradient of the total head depends upon the vorticity, Ñv. The vector (Ñv) is perpendicular to the velocity vector; hence, the gradient of the total head has no component along a streamline. Thus, along a streamline in an incompressible, inviscid, steadyflow,
P rv2
2 gyconstant (10-19)
This is, of course, Bernoulli’s equation, which was discussed in Chapters 6 and 9. If the flow is irrotational and steady, equation (10-18) yields the result that Bernoulli’s equation is
valid throughout theflowfield. An irrotational, steady, incompressibleflow, therefore, has a constant total head throughout theflowfield.3
▶ 10.6
UTILIZATION OF POTENTIAL FLOW
Potentialflow has great utility in engineering for the prediction of pressurefields, forces, and flow rates. In thefield of aerodynamics, for example, potentialflow solutions are used to predict force and moment distributions on wings and other bodies.
An illustration of the determination of the pressure distribution from a potentialflow solution may be obtained from the solution for theflow about a circular cylinder presented in Section 10.3. From the Bernoulli equation
P rv2
2 constant (10-20)
We have deleted the potential energy term in accordance with the original assumption of uniform velocity in thexdirection. At a great distance from the cylinder the pressure isP∞, and the velocity is v∞, so equation (10-20) becomes4
Prv2
2 P∞rv2∞
2 P0 (10-21)
where P0is designated the stagnation pressure(i.e., the pressure at which the velocity is zero). In accordance with equation (10-19), the stagnation pressure is constant throughout thefield in an irrotationalflow. The velocity at the surface of the body isvq 2v∞sinq, thus the surface pressure is
PP0 2rv2∞sin2q (10-22)
A plot of the potentialflow pressure distribution about a cylinder is shown in Figure 10.5.
0 30 60 90 120 150 180
–3 –2 –1 0 1 2
1 2
Figure 10.5 Pressure distribution on a cylinder inan inviscid, incompressible, steadyflow.
3A more general result, Crocco’s theorem, relates the vorticity to the entropy. Thus, it can be shown that a steady, inviscid, irrotationalflow, either compressible or incompressible, is isentropic.
4The stagnation pressure as given in equation (10-21) applies to incompressibleflow only.
10.6 Utilization of Potential Flow ◀ 135
▶ 10.7
POTENTIAL FLOW ANALYSIS—SIMPLE PLANE FLOW CASES
In this section, a number of cases will be considered in which solutions are achieved for two- dimensional, incompressible irrotationalflow. We begin with some very straightforward flow situations.
Case 1.Uniformflow in thexdirection.
For a uniform flow parallel to the x axis, with velocity v∞constant, the stream function and velocity potential relationships are
vxv∞ @Y
@y @f
@x vr0 @Y
@x@f
@y which integrate to yield
Yv∞y fv∞x Case 2.A line source or sink.
A line source, in two dimensions, is aflow that is radially outward from the source, which is the origin in this example. The reverse, or sinkflow, has theflow directed inward.
The source strength is the volumeflow rate per unit depth,Q2prvr:The radial velocity associated with a source is
vr Q 2pr
and the azimuthal velocity is given byvq0. The stream function and velocity potential are evaluated from the expressions
vr Q 2pr1
r
@Y
@q @f
@r vq0 @Y
@r 1 r
@f
@q Integrating these expressions, we obtain for the line source
Y Q 2pq f Q
2plnr
For sinkflow, the sign of the radial velocity is negative (inward), and thus,Qis negative.
The expressions for a line source or sink present a problem atr0, the origin, which is a singular point. At r0, the radial velocity approaches infinity. Physically this is unrealistic, and we use only the concept of line source or sink flow under conditions where the singularity is excluded from consideration.
Case 3. A line vortex.
Vortexflow is that which occurs in a circular fashion around a central point, such as a whirlpool. Afree vortexis one wherefluid particles are irrotational, i.e., they do not rotate as they move in concentric circles about the axis of the vortex. This would be analogous to people sitting in cabins on a ferris wheel. For an irrotationalflow in polar coordinates (see Appendix B), the productrvqmust be constant. The stream function and velocity potential can be written directly,
vr01 r
@Y
@q @f
@r vq K
2pr @Y
@r 1 r
@f
@q which, upon integration, become
Y K 2plnr f K
2pq
whereKis referred to as thevortex strength. WhenKis positive, theflow is observed to be counterclockwise about the vortex center.
▶ 10.8
POTENTIAL FLOW ANALYSIS—SUPERPOSITION
It was shown earlier that both the stream function and the velocity potential satisfy Laplace’s equation for two-dimensional, irrotational, incompressible flow. As Laplace’s equation is linear we can use known solutions to achieve expressions for both Y and f for more complex situations using the principle ofsuperposition. Superposition, simply put, is the process of adding known solutions to achieve another—i.e., ifY1andY2are solutions to Ñ2Y0, then so isY3Y1Y2a solution.
The reader is reminded that the solutions obtained for these very specialized flow conditions are idealizations. They apply for inviscidflow, which is a reasonable approxi- mation for conditions outside the region, near a solid body, where viscous effects are manifested. This region, the boundary layer, will be considered in some depth in Chapter 12.
Some cases will now be considered where the elementary planeflows of the previous section give some interesting and useful results through the process of superposition.
Case 4. The doublet.
A useful case is achieved from considering a source-sink pair on the x axis as the separation distance,2a,approaches zero. Geometrically, we can note that the streamlines and velocity potential lines are circles with centers on theyandxaxes, but with all circles passing through the origin that is a singular point.
The strength of a doublet, designatedl, is defined as thefinite limit of the quantity 2aQ asa®0. For our case, the source is placed on thexaxis at aand the sink is placed on the xaxis ata. The resulting expressions forY andfin polar coordinates are
10.8 Potential Flow Analysis—Superposition ◀ 137
Ylsinq r flcosq
r
Case 5.Flow past a half body—superposition of uniformflow and a source.
The stream function and velocity potentials for uniformflow in thexdirection and for a line source are added together, yielding
YYuniform flowYsource
v∞y Q
2pqv∞rsinq Q 2pq ffuniform flowfsource
v∞x Q
2plnrv∞rcosqQ 2plnr
Case 6.Flow past a cylinder—superposition of uniformflow and a doublet.
As a final illustration of the superposition method, we will consider a case of considerable utility. When the solutions for uniformflow and the doublet are superposed, the result, similar to the past case, defines a streamline pattern inside and around the outside surface of a body. In this case the body is closed and the exteriorflow pattern is that of ideal flow over a cylinder. The expressions forYand fare
YYuniform flowYdoublet
v∞y lsinq
r v∞rsinq lsinq r
v∞r l r
sinq ffuniform flowfdoublet
v∞xlcosq
r v∞rcosqlcosq r
v∞rl r
cosq
It is useful, at this point, to examine the above expressions in more detail. First, for the stream function
Y v∞r l r
sinq
v∞r 1 l=v∞ r2
sinq
where, as we recall,lis the doublet strength. If we chooselsuch that l
v∞a2 wherea is the radius of our cylinder, we obtain
Y r;q v∞rsinq 1 a2 r2
which is the expression used earlier, designated as equation (10-11).