See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324605977

Newtonian analysis of a folded chain drop

Article  in  The Physics Teacher · May 2018

DOI: 10.1119/1.5033873

CITATIONS READS

6 990

1 author:

Carl Mungan
United States Naval Academy
195 PUBLICATIONS   1,939 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Carl Mungan on 19 April 2018.

The user has requested enhancement of the downloaded file.

Newtonian Analysis of a Folded Chain
Drop
Carl E. Mungan, U.S. Naval Academy, Annapolis, MD

C
onsider a chain of length L that hangs in a U shape give caution to any instructor who is tempted to quickly
with end A fixed to a rigid support and free end E re- explain the physics of this system to their students without
leased from rest starting from the same initial height justifying the assumptions and models underlying their ex-
(call it y = 0) as A. Figure 1 sketches the chain after end E has planation.5 One approach to the chain drop problem has been
fallen a distance y. Points O and A are assumed to be close to assume that section DE in Fig. 1 is in free fall so that v(y) =
enough to each other and the chain flexible enough that the vfree(y). For example, a recent paper that makes and then ap-
radius of curvature r at the bottom point C can be taken to be pears to theoretically validate this assumption is by de Sousa
negligibly small (compared to the length of the chain). The and Rodrigues.6 To reemphasize how difficult it can be to sort
problem is to compare the speed of descent v(y) = dy/dt of the out this matter, I challenge the readers to find the flaw in their
free end E of the chain to the speed  of a free- theoretical analysis before they look up my online appendix7
falling point mass that has descended the same distance y. If presenting it. It turns out that making this free fall assump-
v(y) > vfree (y) for all y > 0, then, in a race to fall any arbitrary tion is equivalent to assuming that all of the downward trans-
distance Y (where 0 < Y < L), the chain end E will always beat lational kinetic energy is lost as each successive link moving
a simultaneously released point mass, because the fall time t with speed v at point D in Fig. 1 turns the bottom corner
for E will be shorter than tfree for the point mass, through point C of the fold in the chain and comes to rest at
point B. Presumably most of the kinetic energy lost by this
. (1) bottom-turning link would in this view be dissipated as ther-
mal energy as it inelastically rubs or jerks against the station-
Experimentally, this outcome of the race is observed in a ary link nearest point B of the chain.8 The situation does not
“Veritasium” video.1 seem apparently different than what happens at the bottom
of a chain falling into a heap on a table. In fact, several articles
Related previous O explicitly make such a comparison between the fall of a folded
work A and an unfolded chain.9,10 The rest of the lost downward
This problem has
y translational kinetic energy of individual links is converted
appeared in mechanics E into overall rotational swinging and vibrational oscillations of
textbooks2 dating back the chain seen in video analysis.11–13
to at least 1897. However, Despite the plausibility of these ideas about mechanical
the earliest discussion of energy dissipation and about the free fall of section DE of the
it in a pedagogical journal v a chain, careful comparison between simulations and experi-
is in a pair of letters to the ments reveals that most of the kinetic energy of the falling
editor3 in 1951, with the chain is not lost but is instead “concentrated” into the increas-
first systematic analysis ingly shorter falling section, leading to v(y) > vfree (y), which
appearing a few years is the second school of thought. In fact, to a good approxima-
later.4 But it is only with tion none of the mechanical energy of the system is lost2,4
the publication of Calkin although I am not alone14 in taking issue with authors11 who
and March’s first paper2 D r B consider this conservation of mechanical energy to be easy to
on the topic in 1989 that justify theoretically. Another system that exhibits conserva-
widespread interest led to C tion of mechanical energy is the bungee jumper,15,16 although
a large number of subse- that is less surprising because the cord can be modeled as an
Fig. 1. A hanging chain with fixed
quent articles continuing end A and freely released end E elastic spring.
up to the present day. in the lab frame of reference. The In view of these subtleties associated with energy, there is
There have been two origin O is the release point of merit to developing alternative methods of analyzing the dy-
main schools of thought E. The positive direction for the namics of the falling chain. An elegant Lagrangian approach
displacement y, velocity v = dy/dt,
about the solution to this and acceleration a = dv/dt of the
has been articulated by Wong and Yasui,17 but that is too
problem, with both sides falling section DE of the chain is sophisticated for an introductory course. The present paper
claiming implicitly or downward. Section AB of the chain uses only Newton’s laws to derive the solution to the problem.
even explicitly that their remains at rest. Section BD is a No other article, to the best of my knowledge, fully derives the
semicircle of radius r (greatly exag-
approach is obviously cor- results using forces and accelerations alone at a level of treat-
gerated in size here for clarity) with
rect. This situation should lowest point C. ment accessible to an introductory calculus-based mechanics

298 The Physics Teacher ◆ Vol. 56, May 2018 DOI: 10.1119/1.5033873
course. Only at the end of the present paper is mechanical becomes22
energy discussed, where it is deduced rather than assumed to
TB – TD = mat = ra/2 . (2)
be conserved.
This equation could alternatively be deduced by writing
Theoretical analysis down Newton’s second law in rotational form in terms of the
Sections AB and DE of the chain in Fig. 1 are approximat- torques on, moment of inertia of, and angular acceleration of
ed in this article as starting out and subsequently remaining section BD about point F in Fig. 2. In the limit that the chain
exactly vertical. For a theoretical infinitely flexible chain, that makes a tight turn so that r ➝ 0, it follows from Eq. (2) that
will only be true if the radius of curvature r of section BD is TB < TD. Approximately the same tension, call its magnitude
zero (i.e., if points A and O in Fig. 1 lie on top of each other). TBD, is exerted all the way around section BD of the chain
For a real chain composed of an odd number of links, sec- (just as is true for a rope that does not slip on a pulley if the
tions AB and DE will initially be vertical if points A and O are mass of both the pulley and of the section of the rope on the
separated by a length of exactly one link, as illustrated in Fig. pulley are negligible and there is no friction at the axle of the
11 of Ref. 11. In contrast, for larger horizontal distances be- pulley). Even though point B is stationary while point D is
tween points A and O, sections AB and DE will initially adopt moving downward at speed v in the lab frame, the upward
an approximately catenary shape,18 and photographs of the tension exerted on point B by the chain above it is equal to the
subsequent motion of the chain after releasing end E can be downward tension exerted on point D by the rotating section
found in Refs. 12 and 13. of chain below it, in agreement with a second “Veritasium”
Whereas section DE of the chain falls as a unit with one video23 that follows up on Ref. 1. This downward tension
overall vertically downward speed v (and corresponding force on D in Fig. 1, over and above gravity, which acts both
downward acceleration of magnitude a), the links of the on the chain section DE and on a falling point mass, is what
chain between points B and D do not move in the directions causes the chain end E to outrace the point mass.
tangential to the instantaneous shape of the chain in Fig. 1. To find a quantitative expression for the speed of the free
For example, point C has a vertical velocity component of v/2 end E, consider the
downward in addition to its horizontal velocity component vertical component
of v/2 rightward. It is analogous to the velocities of points on of Newton’s second TBD TBD
the rim of a tire rolling downward without slipping19 along law in the lab frame
line AB: point B then represents the bottom of the tire at rest of reference for this F
relative to the line, point D the top of the tire moving down- semicircular loop of D B
ward at speed v, and point C the front-most point of the tire chain, as illustrated a/2 r
moving downward and rightward with each of those velocity in Fig. 3. In addition ac θ
components equal to half of v. to the upward ten-
It is helpful to sion TBD acting on v /2 C
TD TB analyze the motion of each of its ends, the
mg
the chain in a frame downward gravita-
of reference moving tional force on the
F downward at speed section is mg = rg. Fig. 3. Forces (TBD and mg) and accel-
D B v/2, as in Fig. 2. (This The acceleration of erations (a/2 and ac) that have a net
frame is accelerating each infinitesimal vertical component for chain section BD
of mass m in the lab frame of reference.
vertically downward segment of the loop
That section is rotating about point F
at the time derivative is the sum of three
at with speed v/2. It is semicircular with
a/2 of its speed.) In terms: the tangen- radius r such that any point on the sec-
m this frame co-moving tial acceleration at tion is located at angle varying from
vertically with point (as shown in Fig. – /2 to /2; for example, the lowest
Fig. 2. Forces (TB and TD) and accelera- point C is at angular position = 0.
tion (at) that have a net tangential com-
C, section BD of the 2), the overall verti-
ponent for chain section BD of mass m chain is rotating with cally downward acceleration a/2 of the falling loop, and the
in the vertically co-moving frame. That speed v/2 around the centripetal acceleration ac = (v/2)2/r as the segment rotates at
section makes a semicircular loop cen- stationary center F of speed v/2 around the center point F. To find the net accelera-
tered on point F.
a semicircle of radius tion of the entire section BD, average together the vector ac-
r, like a rope on an ideal pulley.20 Thus its tangential accelera- celerations of each individual segment and then multiply that
tion21 is at at = a/2. The mass of this semicircular section is m average acceleration by the total mass m = r of the section.
= r, where is the linear mass density of the chain. Denote (Equivalently, integrate the product of the vector acceleration
the magnitudes of the vertically upward tension on ends B and infinitesimal mass of each segment over the entire sec-
and D of this section as TB and TD, respectively. The tangential tion.) By symmetry, the average vertical component of the
component of Newton’s second law for section BD of the chain tangential acceleration is zero, as is the average horizontal

The Physics Teacher ◆ Vol. 56, May 2018 299

component of the centripetal acceleration. The average verti-
cal component of the centripetal acceleration is

, (3)

where is the angle from F to any given infinitesimal segment

of the chain relative to the vertical direction, as indicated in
Fig. 3. Assembling all the terms for the
vertical component of Newton’s second O
law for section BD gives y Fig. 5. Speed of the free end of the chain relative to the free fall
speed of a point mass as a function of the distance of fall relative
to the length of the chain.
(4)
E
Again putting r ➝ 0 results in after accounting for the initial condition v(0) = 0 . Thus

v (5) (11)

The left-hand equation predicts that a which is indeed greater than 1 for any y in the range 0 <y < L,
TBD = 0 if = 0. Thus in Hewitt’s race 24 as plotted in Fig. 5. For example, the speed of the free end E
λg(L–y)/2
between a free block and a block at- of the chain is double the free fall speed when y = 6L/7. The
tached to the end of a folded chain, the speed v diverges when y = L as the last bit of the chain “whips”
chained block only wins if the chain has around the bottom turn.4 Likewise the acceleration a, which
mass.5 Also TBD = 0 if v = 0. That is, be- can be found by substituting Eq. (10) into (7), diverges when
fore the release of end E the tension in D y = L. This divergence in v and a does not occur for a real
the chain at the bottom fold is zero, as TBD chain because the last link at the free end has a finite length
expected for a static hanging chain. and mass (rather than continuously decreasing to zero) anal-
Now consider section DE of the Fig. 4. Free-body dia- ogous to the “cracker” at the tip of a whip.25
chain in isolation, as sketched in Fig. gram for section DE Equation (10) is consistent with conservation of mechani-
4. Newton’s second law for it becomes of the chain in the cal energy,26 in which the kinetic energy lost by a link in
lab frame of refer-
ence. In the limit r ➝ moving from point D to B in Fig. 1 is added to the kinetic
TBD + g(L – y) /2 = a (L – y) /2. (6) 0 so that section BD energy of section DE of the chain. As the link turns the corner
in Fig. 1 has negligible at point C, it is pulled back by section DE, thereby slowing
Solve this equation for TBD and substi- length, section DE has the link down to rest and correspondingly speeding up the
length (L–y)/2 decreas-
tute it into the right-hand side of Eq. ing from L/2 to 0, and
falling section of chain. Think of this bottom link as a small
(5) to get section AB has length grappling hook connected to section DE by a very stiff elastic
(L + y)/2 increasing band. To go around the corner, that band must stretch slight-
v2 = 2(L – y)(a – g). (7) from L/2 to L as y ly, thereby slowing down the hook. Just as the hook comes to
increases from 0 to L.
The sum of lengths AB
rest, it attaches without dissipation to fixed point B. The elas-
Separately using the chain rule, the and DE is always L. tic potential energy in the band now pulls on and speeds up
acceleration is the rest of the falling chain, giving it extra kinetic energy.

Acknowledgments
(8) Seth Rittenhouse and later Alpha Wilson proposed that sec-
tion BD of the chain behaves as if it rotates around a falling
so that Eq. (7) becomes pulley. Bob Morse suggested analyzing the torques on that
section. Rich Downey commented on the tension along the
, chain.
(9)
which can be integrated with respect to y as References
1. “Chain Drop Experiment,” YouTube, https://www.youtube.
, (10) com/watch?v=1erU-Cwcl2c.
2. M. G. Calkin and R. H. March, “The dynamics of a falling
chain I,” Am. J. Phys. 57, 154–157 (Feb. 1989).

300 The Physics Teacher ◆ Vol. 56, May 2018

3. J. S. Miller, “An extension of the falling chain problem,” and J. watch?v=X-QFAB0gEtE.
Satterly, “Falling chains,” letters to the editor, Am. J. Phys. 19, 24. P. Hewitt, “Chain drop revisited,” Phys. Teach. 54, 329 (Sept.
383–384 (Sept. 1951). 2016) and 439 (Oct. 2016).
4. W. A. Heywood, H. Hurwitz, and D. Z. Ryan, “Whip effect in a 25. P. Krehl, S. Engemann, and D. Schwenkel, “The puzzle of whip
falling chain,” Am. J. Phys. 23, 279–280 (May 1955). cracking uncovered by a correlation of whip-tip kinematics
5. C. E. Mungan, “The challenge of explaining why,” Phys. Teach. with shock wave emission,” Shock Waves 8, 1–9 (Feb. 1998).
54, 388–389 (Nov. 2016). 26. Conservation of mechanical energy for the falling chain from
6. C. A. de Sousa and V. H. Rodrigues, “Mass redistribution in the instant before release to the instant shown in Fig. 1 is ex-
variable mass systems,” Eur. J. Phys. 25, 41–49 (Jan. 2004). pressed as
7. The appendix is available at TPT Online, http://dx.doi.
org/10.1119/1.5033873, under the Supplemental tab.
8. W. Steiner and H. Troger, “On the equations of motion of the which rearranges into Eq. (10).
folded inextensible string,” Z. Angew. Math. Phys. 46, 960–970
(Nov. 1995). Carl Mungan has been a professor of physics for 18 years in Annapolis,
9. E. Hamm and J.-C. Géminard, “The weight of a falling chain nestled along the Chesapeake Bay.
revisited,” Am. J. Phys. 78, 828–833 (Aug. 2010). U.S. Naval Academy, Physics Department, Annapolis, MD 21042;
10. A. Grewal, P. Johnson, and A. Ruina, “A chain that speeds up, [email protected]
rather than slows, due to collisions: How compression can
cause tension,” Am. J. Phys. 79, 723–729 (July 2011) with a
Supplemental Simulation on p. 981 of the Sept. 2011 issue.  
11. W. Tomaszewski and P. Pieranski, “Dynamics of ropes and
chains I: The fall of the folded chain,” New J. Phys. 7, (45), 1–17
(Feb. 2005).
12. W. Tomaszewski, P. Pieranski, and J.-C. Géminard, “The mo-
tion of a freely falling chain tip,” Am. J. Phys. 74, 776–783 (Sept.
2006).
13. J.-C. Géminard and L. Vanel, “The motion of a freely falling
chain tip: Force measurements,” Am. J. Phys. 76, 541–545 (June
2008).
14. M. Schagerl, A. Steindl, W. Steiner, and H. Troger, “On the par-
adox of the free falling folded chain,” Acta Mech. 125, 155–168
(March 1997).
15. D. Kagan and A. Kott, “The greater-than-g acceleration of a
bungee jumper,” Phys. Teach. 34, 368–373 (Sept. 1996).
16. H. Biezeveld, “The bungee jumper: A comparison of predicted
and measured values,” Phys. Teach. 41, 238–241 (April 2003).
17. C. W. Wong and K. Yasui, “Falling chains,” Am. J. Phys. 74,
490–496 (June 2006).
18. When points A and O in Fig. 1 are far apart, sections AB and
DE will initially be exact catenaries [as in F. Behroozi, “In
praise of the catenary,” Phys. Teach. 56, 214–217 (April 2018)]
only for a uniform string rather than for a chain having discrete
links.
19. See Fig. 2 in T. Pritchett, R. C. Nelson, T. J. Creamer, and B. G.
Oldaker, “Does an ideal wheel really rotate about its instan-
taneous point of contact?” Phys. Teach. 36, 167–170 (March
1998).
20. The top fold in a chain fountain can likewise be thought of as
going around an ideal pulley. See J. Pantaleone, “A quantitative
analysis of the chain fountain,” Am. J. Phys. 85, 414–421 (June
2017).
21. The tangential acceleration of the entire chain in the co-mov-
ing frame is a/2, which is the average in the lab frame of the ac-
celeration a of section DE and the zero acceleration of section
AB.
22. The other three forces and accelerations introduced in connec-
tion with Fig. 3 (namely mg, a/2, and ac) have zero net tangen-
tial component and so do not appear in Eq. (2).
23. “Chain Drop Answer 2,” https://www.youtube.com/

The Physics Teacher ◆ Vol. 56, May 2018 301

 Online Appendix for “Newtonian Analysis of a Folded Chain Drop”

A novel rocket equation approach was used by de Sousa and Rodrigues1 to incorrectly
deduce that the released end of the chain is in freefall such that its acceleration is a = g .
Surprisingly, the specific flaw in the analysis is not pinpointed in a subsequent related paper2 by
one of the authors. Equation (14) of Ref. 1 can be rewritten as
d dM
( M υ ) = Mg + u (A1)
dt dt
where they put u = υ . Here M is the mass of section DE of the chain in Fig. 1 in the main body
of the present article which is equal to λ ( L − y) / 2 . The problem is that de Sousa and Rodrigues
did not account for the semicircular bend in the chain, which is effectively a third subsystem (in
addition to sections AB and DE of the chain). Applying their Eq. (11) with subsystem I taken to
be section BD of the chain (moving downward with average speed u = υ / 2 ) and subsystem II
taken to be section DE of the chain, then Eq. (A1) gives the correct answer. That is, Eq. (A1) is
correct only for u = υ / 2 rather than u = υ , as also noted in Sec. III of Wong and Yasui.3 As a
check, Eq. (11) of Ref. 1 can be used again, with the same subsystem I but this time with
subsystem II taken to be section AB of the chain. Then Eq. (A1) becomes
d dM ʹ
(0) = M ʹg − TA + u (A2)
dt dt
where M ʹ = λ ( L + y ) / 2 is the mass of section AB of the chain which is at rest, explaining the
zero momentum on the left-hand side of Eq. (A2). Here TA is the tension with which the support
(indicated by hash marks in Fig. 1 in the main body of the present article) pulls upward on end A
of the chain, equivalent to what de Sousa and Rodrigues call −T ( x) . Equation (A2) can be
rearranged to find

TA = 12 λ ( L + y) g + 14 λυ 2 (A3)

in agreement with Eq. (13) of Calkin and March,4 in contrast to the erroneous Eq. (16) of de
Sousa and Rodrigues.1
An alternative correct approach is to consider all of the chain from points B to E in Fig. 1 (in
the main body of the present article) to be one system of mass M and downward speed υ. Since
the two forces on this system are Mg downward and tension TBD upward on end B, one finds
d
( M υ ) = Mg − TBD . (A4)
dt
Likewise consider a second system to be section AB of mass M ʹ at rest. The three forces on this
second system are M ʹg downward, TA upward on end A, and TBD downward on end B so that
d
(0) = M ʹg − TA + TBD . (A5)
dt

1
 The reader is invited to check that Eqs. (A4) and (A5) are fully consistent with the equations
presented in the main body of the present article. In particular, the sum of these two equations
implies that the momentum change of the whole chain equals the net external force acting on it.
Methods similar to those presented in this Appendix can be used to find the tension at any
other point of interest along sections AB or DE of the chain as a function of the distance y that
free end E has fallen (in Fig. 1 in the main body of the present article).

1. C.A. de Sousa and V.H. Rodrigues, “Mass redistribution in variable mass systems,” Eur. J.
Phys. 25, 41–49 (Jan. 2004).
2. C.A. de Sousa, P.M. Gordo, and P. Costa, “Falling chains as variable-mass systems:
Theoretical model and experimental analysis,” Eur. J. Phys. 33, 1007–1020 (July 2012).
3. C.W. Wong and K. Yasui, “Falling chains,” Am. J. Phys. 74, 490–496 (June 2006).
4. M.G. Calkin and R.H. March, “The dynamics of a falling chain I,” Am. J. Phys. 57, 154–157
(Feb. 1989).

View publication stats