CE 319F - Laboratory #6

Dimensional Analysis Applied to Drag Force

Objectives
Theory
Part 1: Dimensional Analysis
Part 2: Terminal Velocity
Laboratory Apparatus
Procedures
List of Figures:
Drag coefficients for smooth spheres
Laboratory Apparatus

Objectives
The primary purpose of this laboratory experiment is to see the application of dimensional analysis to obtain
dimensionless parameters that reduce the number of variables needed to represent a physical process. The
physical process in this laboratory experiment is the drag (resistance) force on spheres moving through a
fluid.

In the experiments, the rate at which spheres fall through liquids will be measured. Using dimensional
analysis, the results apply equally well to spheres falling through gases instead of liquids or light spheres
rising through gases or liquids. The results of these experiments and any others should be used only within
the same range of dimensionless variables for which the tests are done.

The Theory section summarizes some concepts of objects moving through fluids. These flows are sometimes
called external flows, since the fluid is outside of the object. Internal flows are flows in pipes and ducts.

Theory
Part 1: Dimensional Analysis
The drag force (FD) on a submerged spherical object is dependent on the diameter of the sphere (D), the
relative velocity between the sphere and the fluid (V), the fluid density (r), and the fluid viscosity (m). In
other words, FD = f(D, V, r, m), where f indicates a functional relationship, which just gives another way of
saying that FD depends on D, V, r, and m. The dimensions (indicated by the square brackets) of each variable
are as follows:

[FD] = [F] = [ML/T2]

D = [L]
 V = [L/T]
r = [M/L3]
m = [FT/L2] = [M/LT]

Since there are 5 dimensional variables and 3 dimensions, there are two dimensionless variables for this
problem. The units can be canceled in steps, beginning with the parameter which has the largest number of
dimensions. In this case, the process will begin with the drag force, which has the largest number of
dimensions when expressed in terms of mass. First divide the force by the density to cancel the mass
dimension:

Since T2 is in the denominator, divide by V2 to cancel the time dimension:

Finally, divide by D2 to cancel the length dimension:

Since the result is dimensionless, the first dimensionless variable is

To be certain that the second dimensionless parameter is independent of the first one, the same process can be
done again beginning with one of the dimensional variables which is not in the first dimensionless parameter
and which has the largest number of dimensions. Thus, start with m since it is the only parameter not in the
first dimensionless term (and since it contains all of the dimensions). Divide by r to cancel the mass
dimension:
Since T is in the denominator, divide by V to cancel the time dimension:

Finally divide by D to cancel the remaining length dimension:

Since this result is dimensionless, the second dimensionless variable is

From the original relation and the two dimensionless parameters,

It can be confirmed that these two parameters are indeed dimensionless by inserting the dimensions for each
term and seeing that all of the dimensions cancel.

There are some dimensionless groups that occur frequently in fluid mechanics and thus have evolved into
standard forms. In this case, P1 is similar to a parameter called a drag coefficient (CD), where

Aproj is the projected area, i.e., the area of the object projected onto a plane perpendicular to the flow so that
Aproj = pD2/4 for a sphere. Comparison with P1 will show that CD = 8P1/p. The general functional
dependence, namely P1 = f(P2), is not changed by constant factors.

P2 is the inverse of a dimensionless parameter called the Reynold's number and abbreviated as Re:

The Reynold's number can be interpreted as the ratio of inertial forces to viscous forces.

The general functional dependence is not changed by constant factors or by combining or inverting
parameters as long as the proper number of independent, dimensionless parameters is maintained. Thus, the
original relation P1 = f(P2) can be written as CD = f(Re). The functional relationship originally represented
by 5 variables is now totally representated by only two variables.

Part 2: Terminal Velocity

The settling of a particle in a fluid can be analyzed by the classic laws of sedimentation formulated by
Newton and Stokes. Newton's law yields the terminal particle velocity, which is the steady velocity that an
object has when its submerged weight is equal to the resistance or drag force exerted by the fluid. At the
terminal velocity (Vt), the sum of the forces is zero so there is no acceleration and the velocity is constant.
An object's submerged weight (Wsubm) is its weight (Ws) minus the buoyant force (FB). The submerged will
be upward if the buoyant force is greater than the weight. The terminal velocity can be found by equating the
submerged weight of an object with the drag force (FD), i.e.,

drag force = submerged weight = weight - buoyant force

FD = Wsubm = Ws - FB

The weight (Ws) of an object with a constant density is

Ws = rsg"s

where rs = the density of the particle, g = the acceleration due to gravity, and "s = the volume of the particle.
For a spherical particle, this volume is

If the object is displacing fluid with a constant density, the buoyant force is

FB = rg"s

where r = the density of the fluid. The submerged weight (Wsubm) is then WB = (rs - r)g"s.

The drag force can be written as

where CD is a drag coefficient. Equating the submerged weight and the drag force yields

where Vt is the terminal velocity. Solving for the velocity yields Newton's law for terminal velocity, which is

The drag coefficient has different values depending on whether the flow surrounding the particle is laminar or
turbulent (i.e., depending on the value of Re). For smooth, spherical particles with Re < 1000, the drag
coefficient can be approximated by

This relationship between the drag coefficient and Reynold's number for smooth spheres is shown below.

Drag Coefficients for Smooth Spheres

For Reynold's numbers less than about 0.5, the flow is called a creeping flow. The accelerations (inertial
forces) of the fluid particles around the object are so small that they are insignificant. The first term
dominates the equation, so CD = 24/Re. Substituting this value into Newton's law yields Stoke's law for
terminal velocity for Re < 0.5, namely,

Laboratory Apparatus
The apparatus consists of three cylindrical tubes, each containing a different liquid. The density and viscosity
of each liquid are known. There are also spheres of known diameter and specific weight.

Laboratory Apparatus

Procedures
1. Release a spherical object of known size and density near the top of one of the tubes.
2. Determine the terminal velocity by letting the object come to its steady velocity and then measuring the
time required for it to fall a known distance.
3. Calculate the submerged weight from Wsubm = (rs - r)g"s.
4. Calculate the drag coefficient from
5. Calculate the Reynolds number from Re = rVtD/m.
6. Plot CD vs. Re on the drag coefficient graph.
7. Repeat this procedure for a few different spheres in different liquids.
8. Note that the results are consistent in terms of CD and Re and are independent of the specific values of
D, V, r, and m.
CE319F (©Kinnas) FALL 1998
Some notes on similitude

The shown (elementary) force vectors are those acting on the elementary fluid cubes (fluid particles),
also shown on the graphs. The pressure force vector (superscript p) is due to the pressures applying on
all sides of the fluid cube, the viscous force vector (superscript v) is due to the shear stresses acting on
all sides of the fluid cube, the gravity force vector (superscript g) is the weight of the cube, and the
inertial force vector (superscript i) is the product of the mass of the fluid cube times the fluid
acceleration vector.

Note 1: The shown fluid cubes are at the same relative location with respect to the body (e.g. building)
and are also scaled by the same scale ratio.

Note 2: As shown in the graph, due to Newton's law applying on each fluid cube, the vector sum of the
pressure, the viscous, and the gravity forces, equals the inertial force vector. In other words, the
pressure, the viscous, the gravity, and the inertial force vectors form a closed polygon!

In order to have similarity of the flow patterns between the model and the prototype (also called Kinematic
Similarity) we must have Dynamic Similarity (or similarity of the shown force polygons):
where the subscript ``p'' stands for prototype and ``m'' for model. The superscript ``v'' stands for viscous, ``p''
for pressure, ``g'' for gravity, and ``i'' for inertial. The inertial force is defined as:

where M is the mass of the fluid particle and a the acceleration.

Due to similarity the following proportionality relations can be written:

where (Delta p) the difference of pressure from the ambient pressure

Using the above equations we can get:

from equating the first fraction with the last fraction in equation (1)

or

where Re is the Reynolds number

from equating the third fraction with the last fraction in equation (1)

or

where Fr is the Froude number

 from equating the second fraction with the last fraction in equation (1)

or

where CP is the pressure coefficient

As mentioned in class it is not feasible to enforce equality of the Reynolds and the Froude numbers at the
same time. Thus, depending on the application we can only force equality of the Reynolds number (internal
flow or external flow in unbounded fluid) or of the Froude number (flow of body through or under free
surface, flow over a spill-way). The equality of the pressure coefficients results AUTOMATICALLY from the
equality of Re or Fr numbers, and thus does not need to be enforced.

So, once the dynamic similarity has been enforced we can:

relate the pressure in the prototype to that (measured) in the model via the equality of the pressure
coefficients (see Ex. 8.6 of the textbook).
relate the total force on the prototype (Fp) to the total force (measured) on the model (Fm) by using the
dynamic similarity relations, e.g. in a case where only the Reynolds number is involved:

You can use any of the fractions in the above equation to determine Fp/Fm. In Ex. 8.7 the textbook
uses (together with the fact that CPm=CPp)

In the solution of Ex. 8.7 which was presented in class I used:

We could have also used:

All of the above choices will give us the same answer for Fp/Fm, due to dynamic similarity (which has
been enforced via equality of Reynolds numbers in the case of Ex. 8.7).

Spyros A. Kinnas