You are on page 1of 7

Home Search Collections Journals About Contact us My IOPscience

Classic Bernoulli’s principle derivation and its working hypotheses

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2016 Phys. Educ. 51 045005

(http://iopscience.iop.org/0031-9120/51/4/045005)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 144.122.201.150
This content was downloaded on 11/05/2016 at 04:59

Please note that terms and conditions apply.


Papers
Phys. Educ. 51 (2016) 045005 (6pp) iopscience.org/ped

Classic Bernoulli’s principle


derivation and its working
hypotheses
Edson R Marciotto
Department of Physics, Universidade Federal de Santa Catarina, 88040900,
­Campus ­Trindade, Florianópolis, Brazil

E-mail: e.r.marciotto@gmail.com

Abstract
The Bernoulli’s principle states that the quantity p + ρgz + ρv /2 must be
2

conserved in a streamtube if some conditions are matched, namely: steady


and irrotational flow of an inviscid and incompressible fluid. In most physics
textbooks this result is demonstrated invoking the energy conservation of a
fluid material volume at two different positions of a pipe whose cross-section
and height vary along its way. Although the final result is correct the right
justifications presented in textbooks are usually unclear or absent. The main
problem rests on the work done by pressure, which are not found to be fully
justified via free-body diagrams as depicted in many general physics textbooks,
not to mention plenty of videos on YouTube that incur in similar omissions.
In this article I will discuss this issue and how it is solved without resorting to
alternative demonstrations. In addition, I discuss the needs of the assumptions
to get the Bernoulli’s principle in a way viable to introductory physics courses.

1. Introduction would be accounted for in terms of Newtonian


It is not from recent times that the famous mechanics as
Bernoulli’s principle is an object of confu­
sion [1, 7]. For example, a fallacious argument Wpressure = ∆K + ∆U,
(1)
about its application to wing lift was clarified
by Babinsky (2003) [2]. The concept of pres­ where Wpressure is the work done by pressure on a
sure energy seems to be the most confused in fluid particle, and ∆K and ∆U are the kinetic and
old physics textbooks until its overall refuta­ potential energy variation, respectively.
tion [5, 12], and the principle derivation itself The derivation usually found in general phys­
has also been an object of debate [10]. However, ics textbooks [3, 8, 11]1 utilize the layout pre­
herein I will bring some unclear statements on sented in figure 1, that I will call pipe-approach.
the Bernoulli’ principle still present in modern Such a sketch is so popular in the physics and
general physics textbooks. engineering communities that it turns out to be the
Most of them present the classic derivation of icon for the fluid mechanics subject in Wikipedia
Bernoulli’s equation by computing the work done [13], from which figure 1 is a redraw.
by pressure forces implying a variation of the
kinetic and the potential energy. This approach 1
Including their new editions by other authors.

0031-9120/16/045005+6$33.00 1 © 2016 IOP Publishing Ltd


E R Marciotto
The stated assumptions in Bernoulli’s princi­
ple are (i) the fluid is assumed to be inviscid and
incompressible, and (ii) the flow is assumed to
be stationary (steady) and irrotational. Then, the
argumentation development follows like this;
From equation (1), the variation of mechani­
cal energy K  +  U  =  mv2/2  +  mgz is caused by
unbalanced forces on the fluid particle due to the
pressure on each surface A and B, which are writ­
ten as pA AA and −pB AB. By applying the mechani­
cal energy conservation and incompressibility, Figure 1.  Layout of the pipe-approach to demonstrate
and doing the algebra, one gets the Bernoulli’s the Bernoulli’s principle.
equation.
Specialised textbooks on fluid mechanics do
derive the Bernoulli’s principle from the Euler could not be −mg(zB − z A) because the ­material
equation (see the appendix for such a derivation). volume would have been displaced just sA (or sB)
The criticism addressed in this paper is only to and the height variation of the center of mass
general physics textbooks, usually employed in would be actually much less than zB − z A.
the beginning of college studies. We still could imagine that there are two
material volumes. One at A displaces sA and the
other at B displaces sB. Nonetheless, a major
2.  Main issues found problem arises here as well. If one is studying
Many doubts arise from the argumentation in the a particle motion via Newton’s law, the forces
last section. The most problematic issue is cer­ must be applied to the very same particle, or
tainly the pressure term derivation. In general, in our pipe-approach, to the same material vol­
though the final result is alright, the correct physi­ ume. Therefore, it is not possible to talk about
cal justifications are unclear, absent, and some­ force balance in this case because pA AA and pB AB
times even incorrect. would be applied on distinguished bodies! It can
be argued that what happens between the front
portion of region A and the rear portion of region
2.1.  The pipe-approach derivation
B is unimportant, since the flow is assumed to
The shadowed areas in figure 1 refer to the very be stationary [6]. In spite of this being true, as
same material volume2, which has been displaced will be shown next section, there is no revealing
from one position to another. Thus, the variation explanation on how stationary hypothesis works
of potential energy is −mg(zB − z A) and the varia­ out the problem.
tion of kinetic energy is mv2B/2 − mv2A/2. However,
assuming that the material volume was displaced 2.2.  The lack of discussion of the working
between those positions, the force on each surface hypothesis
of the material volume would have continuously
The assumptions needed to derive the Bernoulli’s
changed as the material volume was displaced.
principle are poorly explored in general phys­
Therefore, we cannot apply the tacit hypothesis
ics textbooks. Students may have a feel that all
of constant force; the force is constant only inside
assumptions, but energy conservation, are good
the constant-radius pipe region.
for nothing. The points of criticisms are:
Perhaps the material volume was the volume
within the portion from the rear at A to the rear at B. (i) it is neither clear what the needed hypotheses
If that were the case, then the work done by gravity to get the result are nor how they imply the
final result; compressibility and irrotation­
2
A material volume is an amount of fluid that conserves it allity is the most unclear hypothesis, and
mass throughout the flow. If the fluid is incompressible then stationarity is poorly discussed [6];
its volume remains constant, although it can change in shape.
This corresponds to the so-called Lagrangian description of (ii) there is no clear definition of what the
a flow. material volume (or fluid particle) which

July 2016 2 P h y s i c s E d u c at i o n
Classic Bernoulli’s principle derivation and its working hypotheses
the forces are applied on is. In fact, some
traditional textbooks imply that there are two
material volumes, one at position A and other
at B. It is not clearly stated that the material
volume flows from A to B, especially when
the pressure term is evaluated. In this case, it
is implied as the opposite; that the whole fluid
from A to B is a single material volume on Figure 2.  Free-body diagram for pressure forces: (a)
which the pressure forces are applied; the very same material volume at two consecutive time-
(iii) the clarification of how a stationary flow steps, k and k  +  1, (b) unbalanced forces on the time
can have different speeds at different step k, and (c) unbalanced forces on time-step k  +  1. By
invoking the stationary hypothesis we can assume that
points is often omitted. This is an issue in for any time-step there is a fluid particle similar to the
some textbooks that define stationarity in studied material volume before and ahead of it. As the
a misleading way. In fact, we can find the material volume moves the pipe changes its gauge, so
association of stationary flow with smooth that the area of the interface between two consecutive
flow, which is not correct [9]. One can material volumes is not exactly the same, as well as its
thickness in that discretised approximation. When the
actually have a non-stationary smooth flow. number of time-steps goes to infinity, the areas Ak+1
For instance, one can figure out water in a and Ak+2 go to the same value, as do the thicknesses
small diameter pipe being pulled by gravity δsk and δsk + 1.
from v0  =  0 to some low speed, say, 0.1
m s−1. In this range no turbulence would
likely be developed, thereby the flow would force to take the material volume from A to B
remain smooth even though non-stationary. must be pA AA − pB AB. Thus, the problem is how
In another textbook [8], it is stated that a to get that pressure work for a material volume
stationary flow is that in which the global that changes its cross-section area continuously
configuration does not change with time. In from A to B, implying a variable force.
both cases the definitions are not helpful as In order to solve this difficulty we will con­
working hypotheses; sider a material volume of mass m and a finite

(iv) the final result obtained from the pipe- volume VA initially at position A being displaced
approach is stated to be held just for a to the final position B, where it gets a finite vol­
streamline, unless the flow be irrotational. ume VB. The incompressibility hypothesis allows
Nonetheless, the discussion commonly us to write VA = VB. The free-body diagram is as
found (if any) provides no clue for that sketched in figure 2. Let us assume that the mat­
understanding. erial volume is displaced from its initial position
only by a small distance δs1. The force on the
material volume on the rear-surface is pA AA and
3.  Revisiting Bernoulli’s principle on the front-surface is −p1 A1, with a minus sig­
3.1.  Pipe-approach derivation nal meaning F1 is against the displacement. The
work done over the material volume due to this
The inconsistencies of the pipe-approach demon­ small displacement is δW1 = ( pA AA − p1 A1)δs1.
stration vanishes considering that there is just one At that new position, the forces on the material
material volume and that it is displaced from A to volume are p1 A1 on the rear-surface and −p2 A2 on
B. Among the three situations discussed before, the front-surface, which causes the displacement
this is the only one that provides the variation of δs2, doing a small work δW2 = ( p1 A1 − p2 A2 )δs2.
kinetic and potential energy, m /2(v2B − v2A) and As a consequence of incompressibility, this would
−mg(zB − z A), respectively, as required in the imply that the displacement steps δsk are varying
pipe-approach demonstration of Bernoulli’s prin­ along the pipe. Based on this construction, we can
ciple. Hence, the variation of mechanical energy split the total work for displacing the elementary
of the fluid particle does not seems to be the issue volume from A to B into the contribution of many
at all. Yet in this case, the work done by pressure elementary works:

July 2016 3 P h y s i c s E d u c at i o n
E R Marciotto

δW1 = ( pA AA − p1 A1)δs1 common language of Fluid Mechanics, a flow is


stationary if the local vector acceleration is zero
 
δWk = ( pk Ak − pk + 1 Ak + 1)δsk at a given point. This is why it is possible for a

fluid particle to change its velocity from one point
  to another with the flow still being stationary.
δWN = ( pN AN − pB AB )δsN (2) If we measure the velocity at points A and B, they
Without loss of generality que can assume would be different but constant at each point (see
δsk = δsk + 1 = δs. Therefore, the total work to appendix B for details). Now, for a non-stationary
take an elementary material volume from A to B flow, the mechanical energy varies with time and
is given by its piecewise sum: the energy conservation theorem (equation (1)
cannot be applied. In addition, the fluid obviously
δWtotal = δW1 + δW2 +  + δWN − 1 + δWN should be inviscid, otherwise the fluid particle
= ( pA AA − p1A1)δs + ( p1A1 − p2 A2 )δs +  would loose mechanical energy along the way from
  + ( pN − 1AN − 1 − pN AN )δs + ( pN AN − pB AB )δs A to B due to the friction, in the very same way the
≈ pA AA δs − pB AB δs mechanical energy of a brick is not conserved when
= pA δVA − pB δVB. (3) it is displaced on a rough inclined plane.
In the pipe-approach derivation found in text­
Wherever the pipe has a constant area we can
books one needs the incompressibility assump­
replace δVA and δVB for a finite volume VA and VB,
tion to write VA = VB so that the volume cancels
respectively:
out and one obtains the Bernoulli’s equation
Wtotal = pA VA − pB VB.
(4) in the form pA + ρ v2A/2 + ρgz A = pB + ρ v2B/2 + ρgzB.
This result shows that the work done by pressure Nonetheless, the form pA VA + ρAVAv2A/2 + ρAVAgz A =
to displace a fluid particle of mass m from posi­ pB VB + ρBVB v2B/2 + ρBVBgzB is also a statement of
tion A to B is equivalent to the difference of pres­ energy conservation. Therefore, incompressibil­
sure times volume of the fluid particle at those ity is not a necessary condition. That is why we
points. The procedure shown above is usually the can apply Bernoulli’s principles to gases, which
missing point when demonstrating Bernoulli’s are relatively compressible. However, the density
principle using this approach. of a gas should change only according to the local
Note that the force on the surface ahead of pressure changes. A flow for which ρ = ρ( p)
the rear particle (or material volume) has the is called barotropic. To clarify this point let us
same magnitude and opposite sense of the force consider the ideal gas state equation in the form
on the rear surface of the particle ahead (figure 2). p = ρR∗T , where R* is a constant for a part­icular
In other words, these forces are action–reaction gas3. If the fluid density changes as a function of
pairs. It is important to warn the student that in the pressure only, there will be no variation in
equation (3) one is not cancelling action–reaction the internal energy of the fluid from point A to B
pairs. What is actually done is to sum elementary (T = constant).
works and just using the fact that forces at a same Finally, the irrotationality is needed because it
material volume interface are equal and oppo­ guarantees that a velocity profile of a one-dimen­
site (action–reaction pairs) to simplify the result. sional flow is planar and thus the flow speed is the
It also should be clear that each elementary work same on a given cross-section. In order to better
occurs at different time-steps (and consequently understand this statement, let us first see how to
different positions) for a given material volume, describe an irrotational flow without resorting to
which is always finite in volume. What becomes partial derivatives. A flow is irrotational at a given
infinitesimal is the volume swapped by the area point (the choice of the point is important) if there
by the displacement from step k to k  +  1. is no net circulation of fluid at that point. This may
be checked by placing at that point a small dented
wheel, which will not turn around. We can speak of
3.2.  Comments on the working hypotheses
3
R*  =  R/M (M is the molecular mass) is not a universal con­
A flow is stationary if its vector velocity does not stant. Instead it depends on the gas or gases in consideration.
change with time at a given point. Using a more For dry air its value is approximately 287 J kg−1 K−1.

July 2016 4 P h y s i c s E d u c at i o n
Classic Bernoulli’s principle derivation and its working hypotheses

∂v 1
+ v ⋅ ∇ v = − ∇p + f.
(A.1)
∂t ρ
If f is a conservative force (per unit volume), we
have f = −∇Φ and then equation  (A.1) can be
written as

∂v 1
+ v ⋅ ∇v = − ∇p − ∇Φ.
(A.2)
∂t ρ
Figure 3. One-dimensional flow profile in a wind 1
tunnel. Although the flow takes place along straight lines Now, using the vector identity v ⋅ ∇v = 2 ∇v 2 −
there is indeed a net circulation. If a dented wheel were v × ∇ × v, we obtain
placed in such a flow it would turn around clockwise. At
(A) the spinning would be faster than at (B). ∂v ⎛1 dp ⎞
+ ∇ ⎜ v2 + Φ + ⎟ = v × ∇ × v.

circulation even in a unidirectional flow. Consider ∂t ⎝2 ρ ⎠

for example, a typical flow profile in a wind tunnel B :Bernoulli function
 (A.3)
as shown in figure 3. Although the flow occurs at
only one direction, there is a net circulation in the As mentioned before, a barotropic fluid is
sense that if we place a dented wheel, it would turn that in which the density is a function of pressure
around with a speed depending on the position in only, that is, ρ = ρ( p). A useful mathematical
the vertical direction. In a unidirectional flow the 1
property of barotropic fluid is that ρ ∇p = ∇ ∫ ρ
dp
rotationality can be reduced to the presence of a
always holds. In order to show this, we will define
velocity gradient, or more simply to a vertical var­
the function dF = (1/ρ )dp. Then
iation in this case. As a consequence, a rotational
and one-dimensional flow has it speed changing dp
according to its vertical position. In this condi­ F=
(A.4)
ρ
. ∫
tion the Bernoulli’s principle cannot obviously
be held for the whole cross-section pipe. Instead, On the other hand,
any streamline has its own Bernoulli’s constant.
Furthermore, an irrotational one-dimensional flow 1 1
∇F ⋅ dr = dF = dp = ∇p ⋅ dr,
(A.5)
is necessarily planar4. ρ ρ
or
Appendix A.  Bernoulli’s principle
1
formally derived ∇F = ∇p,
(A.6)
ρ
In more advanced textbooks on fluid mechanics
the Bernoulli’s principle is usually derived from from which we find the required relation by plac­
the Euler equation, which already assumes a zero- ing equation (A.4) into equation (A.6).
viscosity fluid, but not incompressibility. Here I For a stationary flow we have ∇B = v × ∇ × v.
will provide a formal derivation of the Bernoulli’s From this we can conclude that v × ∇ × v is nor­
principle following the approach by Kundu and mal to the level surfaces of B, so that they contain
Cohen [4] with the aim of helping instructors to both streamlines and vorticity lines, hence
see how the working hypotheses appear to sim­
plify the equations. 1 2
v +Φ+ ∫ dρp = constant,  along any 
The momentum per unit mass of an invis­ 2
cid fluid moving under the action of the pressure given streamline. (A.7)
gradient and a body force (e.g. gravity) can be
described by: B is also constant along any vorticity line. If
4
besides stationary, the flow is also irrotational,
In effect, ∇ × v = 0 is equivalent to ∂vz /∂x − ∂vx /∂z = 0 in
a two-dimensional flow. If the flow is one-dimensional, say then ∇ × v = 0 and B is constant everywhere in
vz  =  0, then vx(z)  =  constant (flat profile). the flow.

July 2016 5 P h y s i c s E d u c at i o n
E R Marciotto
Appendix B.  Changing the fluid particle References
velocity in a stationary flow [1] Awaiti K M 1997 Answer to question #44.
Bernoulli’s principle Am. J. Phys. 65 271
Saying that the velocity of a fluid particle (= mat­ [2] Babinsky H 2003 How do wings work? Phys.
erial volume) can change in a stationary flow can Educ. 38 497–503
seem confusing to the student. In this appendix [3] Halliday D, Resnick R and Walker J 2009
some simple examples are provided of how this Fundamentos de Física: Gravitaç, Ondas
e Termodinâmica vol 2, 9th edn (Rio de
is possible in an accessible way to a student at the Janeiro: Editora LTC)
beginning of a physics course. We just need the [4] Kundu P K and Cohen I M 2002 Fluid
precise definition of stationary flow (beginning Mechanics (New York: Academic)
of section 3.2) and mass conservation law for an pp 110–1
incompressible fluid, i.e. A1u1 = A2 u2. [5] Lindsay G A 1952 Pressure energy and
Bernoulli’s principle Am. J. Phys. 20 86
Let us assume a fluid particle is flowing [6] Nussenzveig H M 1987 Curso de Física Básica
in a wind tunnel with a conic contraction (say, vol 2 (São Paulo: Edgar Blücher)
A2 < A1). Let us set the fan rotation rate constant [7] Ohanian H C 1996 Question #44. Bernoulli
such that the flow can be assumed stationary principle Am. J. Phys. 64 681
(= constant velocity at a given point). As the fluid [8] Sears F W, Zemansky M W, Young H D and
Freedman R A 2008 Física vol 2, 12nd edn
particle passes by the contraction it is accelerated (São Paulo: Pearson)
because more fluid has to pass by per unit area. [9] Serway R and Jewett J W 2004 Physics for
This is called advective acceleration and is related Scientists and Engineers 6th edn (Belmont,
to the surrounds of the flow. Thus, the fluid par­ CA: Thomson Brooks/Cole)
ticle velocity at a position, say r = r2, is greater [10] Sleator W W 1949 Check and proofs of the
Bernoulli equation Am. J. Phys. 17 110
than at position r = r1. However, every particle [11] Tipler P A and Mosca G 2009 Física Para
passing by r1 has the same vector velocity as any Cientistas e Engenheiros vol 2, 6th edn
other particle at that position would have, and [12] Van Lear G A Jr 1934 Pressure energy—a
analogously for the position r2. Whether we can misconception Am. J. Phys. 2 99
precisely control the fan rotation rate, we could [13] Fluid mechanics Wikipedia (https://en.wikipedia.
org/wiki/Fluid_mechanics) (accessed 2 May
do something more interesting, such as decrease 2016)
the fan rotation rate in such a way that the local
negative acceleration (the flow slowing down
locally) would be exactly compensated by the
advective acceleration due to the contraction. The Edson R Marciotto received his PhD
result would be a fluid particle travelling through in meteorology at Universidade de
the contraction with constant vector velocity even São Paulo. He is assistant professor
at Universidade Federal de Santa
though the overall flow outside the contraction is Catarina, Florianópolis, Brazil, where
non-stationary (decreasing velocity). he teaches physics, micrometeorology
Received 3 November 2015 and meteorological instrumentation.
Accepted for publication 24 March 2016
doi:10.1088/0031-9120/51/4/045005

July 2016 6 P h y s i c s E d u c at i o n

You might also like