Hourglass Redux

A well-known physics problem asks whether the weight of an hourglass sitting on a scale would be the same whether all of the sand is at rest, or whether it is flowing. Usually, the answer given is yes, because of the exact balance between the "missing weight" of the sand in free fall, and the force of the sand hitting the bottom.

This well-known physics result is wrong .

The hourglass above has a total mass M M , constructed with a constant cross sectional area A A on the top and bottom, and a very thin section connecting them which allows the sand with density ρ \rho to flow through at a constant rate of mass flow q q . The total amount of sand fills the bottom of the hourglass once all the sand has fallen.

What is the weight of the hourglass at steady state when the sand is flowing?

Note: You may assume that the mass of the falling sand at any given time is negligible compared to the masses at rest above and below, and that once the sand lands, it rapidly spreads out onto a flat surface.

M g + 4 q 2 ρ A Mg + \frac{4q^2}{\rho A} M g + 2 q 2 ρ A Mg + \frac{2q^2}{\rho A} M ( g + 2 q ρ A h ) M \left( g + \frac{2q}{\rho Ah}\right) M g + q 2 ρ A Mg + \frac{q^2}{\rho A}

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

4 solutions

Arjen Vreugdenhil
May 20, 2018

Let m 1 m_1 be the mass of sand in the top and m 2 m_2 the mass of sand in the bottom of the hourglass. Then the height of the sand in each part is y i = m i / ρ A y_i = m_i/\rho A . We calculate the moment m y c m my_{cm} for each section: part mass y c m m y c m top m 1 L + m 1 / 2 ρ A m 1 ( L + m 1 / 2 ρ A ) bottom m 2 m 2 / 2 ρ A m 2 2 / 2 ρ A \begin{array}{rccc} \text{part} & \text{mass} & y_{cm} & my_{cm} \\ \hline \text{top} & m_1 & L + m_1/2\rho A & m_1(L + m_1/2\rho A) \\ \text{bottom} & m_2 & m_2/2\rho A & m_2^2/2\rho A \\ \hline \end{array} Thus for the total amount of sand (ignoring the sand that is flowing) the center of mass is given by M y c m = m 1 L + m 1 2 + m 2 2 2 ρ A . My_{cm} = m_1L + \frac{m_1^2 + m_2^2}{2\rho A}. The time derivative describes the speed of the center of mass. Note that d m 1 / d t = q dm_1/dt = -q and d m 2 / d t = q dm_2/dt = q : M v c m = q L + q ( m 2 m 1 ) ρ A . Mv_{cm} = -qL + \frac{q(m_2 - m_1)}{\rho A}. Taking the derivative again shows that the (downward) acceleration of the center of mass satisfied M a c m = 2 q 2 ρ A , Ma_{cm} = \frac{2q^2}{\rho A}, and this is the additional amount shown by the scale.


What about the sand in the middle? Note, first of all, that d m 2 / d t > q dm_2/dt > q because the amount of sand in the middle part of the container decreases. To analyze this in more detail, model the falling sand as free fall motion from rest. In reality, the sand has an initial speed of q / ρ A q/\rho A , but this we assume to be small enough to ignore relative to the final speed of the sand.

The sand falls over a distance h y h - y ; the time needed to do so is Δ t = 2 ( h y ) / g \Delta t = \sqrt{2(h-y)/g} . This means that the amount of falling sand is m 3 = q Δ t = q 2 ( h y ) g . m_3 = q\Delta t = q\sqrt{\frac{2(h-y)}g}. The average speed of the falling sand is half of its final speed, v ˉ 3 = 1 2 g Δ t = g ( h y ) 2 . \bar v_3 = \tfrac 12 g \Delta t = \sqrt{\frac{g(h-y)}2}. Now consider again the three parts of the sand and their masses and average speeds: part mass v ˉ m v ˉ top m 1 q / ρ A m 1 q / ρ A bottom m 2 0 0 falling q 2 ( h y ) / g g ( h y ) / 2 q ( h y ) \begin{array}{rccc} \text{part} & \text{mass} & \bar v & m\bar v \\ \hline \text{top} & m_1 & q/\rho A & m_1q/\rho A \\ \text{bottom} & m_2 & 0 & 0 \\ \text{falling} & q\sqrt{2(h-y)/g} & \sqrt{g(h-y)/2} & q(h-y) \\ \hline \end{array} Now the total mass of sand is constant: m 1 + m 2 + q 2 ( h y ) / g = M m_1 + m_2 + q\sqrt{2(h-y)/g} = M . Take the time derivative and use m 2 = ρ A y m_2 = \rho A y : d m 1 d t + d d t ρ A y + d d t ( q 2 ( h y ) g ) = 0 ; \frac{dm_1}{dt} + \frac{d}{dt}\rho A y + \frac{d}{dt}\left(q\sqrt{\frac{2(h-y)}g}\right) = 0; q + ρ A d y d t q 2 ( h y ) g d y d t = 0 ; -q + \rho A \frac{dy}{dt} - \frac{q}{\sqrt{2(h-y)g}}\frac{dy}{dt} = 0; d y d t = q ρ A q / 2 ( h y ) g . \frac{dy}{dt} = \frac{q}{\rho A - q/\sqrt{2(h-y)g}}.

Now the speed of the sand's center of mass is M v c m = m v ˉ = m 1 q ρ A + q ( h y ) , Mv_{cm} = \sum m\bar v = \frac{m_1q}{\rho A} + q(h-y), and the time derivative is M a c m = q 2 ρ A q d y d t = q 2 ρ A ( 1 + 1 1 q / ρ A 2 ( h y ) g ) = q 2 ρ A ( 1 + 1 1 q 2 / ρ A m 3 g ) . \begin{aligned} Ma_{cm} & = \frac{q^2}{\rho A} - q\frac{dy}{dt} \\ & = \frac{q^2}{\rho A}\left(1 + \frac{1}{1 - q/\rho A \sqrt{2(h-y)g}}\right) & = \frac{q^2}{\rho A}\left(1 + \frac{1}{1 - q^2/\rho A m_3 g}\right). \end{aligned} Compare the expression in brackets to the constant 2 in the original solution. This constant is justified if q 2 < < ρ A m 3 g q^2 << \rho A m_3 g , i.e. if the rate at which the sand falls is slow compared to the size of the container and the amount of falling sand.

Moderator note:

Buried in the comments to the other answer, but worth bringing out here: this link includes a video demonstrating the problem.

Note this problem essentially is 1 out of a series of 3, spanning through all the levels:

Science problem from Basic

Science problem from Intermediate

Science problem from Advanced

Now suppose we use the approach of my second solution but ignore the falling sand. Then the total momentum of the sand is equal to that of the downward movement in the top container, M v c m = m 1 q ρ A Mv_{cm} = \frac{m_1q}{\rho A} and the force is simply M a c m = q 2 ρ A . Ma_{cm} = \frac{q^2}{\rho A}. Why is this precisely half of the approximation in the first part?

In this solution, we forget to account for the transport of sand from the top to the bottom container, which is modeled to be instantaneous (since we ignore the movement of the falling sand). This "teletransport" of mass at rate q q over distance L y L - y contributes an additional momentum term: M v c m = m 1 q ρ A + q ( L y ) = ( m 1 m 2 ) q ρ A + q L , Mv_{cm} = \frac{m_1q}{\rho A} + q(L - y) = \frac{(m_1 - m_2)q}{\rho A} + qL, which gives the more correct M a c m = 2 q 2 ρ A . Ma_{cm} = \frac{2q^2}{\rho A}.

Arjen Vreugdenhil - 3 years ago

But why you sometimes use L and sometimes use h won't the height of the top part be h+m_2/ρA? That's what the image suggest, or you are assuming the height of the very thin section is 0? I know that it cancels out when deriving for the second time and so the result is the same, neither h nor L will appear on the weight.

Pau Cantos - 3 years ago

Log in to reply

Sorry, when I write L L it should be h h .

Arjen Vreugdenhil - 3 years ago
Blake Farrow Staff
May 16, 2018

The difficult step required to find the difference between the "resting" mass of the hourglass and the steady state "flowing" mass is identifying the additional force that would be present during the flow.

This force is due to the motion of the center of mass of the sand. The center of mass is shifting from the top of the hourglass to the bottom. If this motion of the sand's center of mass is accelerating , then it is causing a force equal to the product of the acceleration and the total mass of the sand.

F f l o w = m s a n d d 2 y c o m d t 2 F_{flow} = m_{sand} \frac{d^2 y_{com}}{d t^2}

First we need to find an expression for the hourglass's center of mass. Let's start with the moment shown in the problem statement:

We can use the equation for the center of mass of two particles:

y c o m = y t o p m t o p + y b o t t o m m b o t t o m m s a n d y_{com} = \frac{y_{top} m_{top} + y_{bottom} m_{bottom}}{m_{sand}}

and we can treat the top and bottom chunks of sand vertically centered at their halfway point:

y c o m = ( h + ( L y ) 2 ) m t o p + ( y 2 ) m b o t t o m m s a n d y_{com} = \frac{\left( h+ \frac{(L-y)}{2} \right) m_{top} + (\frac{y}{2}) m_{bottom}}{m_{sand}}

The mass of the top, bottom, and total portions of sand can be calculated by multiplying the cross-sectional area by the density and height of each section. At the point in the image, the top section has a filled height of ( L y ) (L-y) , while the bottom is filled to y y . The total volume of sand is a full section, which has height L L .

y c o m = ( h + ( L y ) 2 ) ( H y ) A ρ + ( y 2 ) y A ρ L A ρ y_{com} = \frac{\left( h+ \frac{(L-y)}{2} \right) (H-y) A \rho + (\frac{y}{2}) y A \rho}{L A \rho}

This can be simplified and results in a quadratic, linear, and constant term with respect to the height y y :

y c o m = ( L 2 + h ) ( 1 + h L ) y + y 2 L y_{com} = \left( \frac{L}{2} + h \right) - \left( 1 + \frac{h}{L} \right) y + \frac{y^2}{L}

Equipped with this expression for the center of mass, we need to express the filled height y y as a function of time t t . We know that the mass flow rate is q q , which tells us that the total mass at time t t is q t qt . Using the density, we can find the volume as a function of time v ( t ) v(t) , which is the filled height times the cross sectional area.

v ( t ) = m ρ v ( t ) = q t ρ y ( t ) A = q t ρ y ( t ) = q t ρ A \begin{aligned} v(t) &= \frac{m}{\rho} \\ v(t) &= \frac{qt}{\rho}\\ y(t)A &= \frac{qt}{\rho}\\ y(t) &= \frac{qt}{\rho A} \end{aligned}

This expression can be substituted into our center of mass expression to find the center of mass of the falling sand:

y c o m ( t ) = ( L 2 + h ) ( 1 + h L ) q t ρ A + q 2 t 2 ρ 2 A 2 L y_{com}(t) = \left( \frac{L}{2} + h \right) - \left( 1 + \frac{h}{L} \right) \frac{qt}{\rho A} + \frac{q^2 t^2 }{\rho^2 A^2 L}

To find the acceleration of the center of mass, we take the double derivate with respect to time. We only need to concern ourselves with the quadratic term:

a c o m = d 2 d t 2 q 2 t 2 ρ 2 A 2 L = 2 q 2 ρ 2 A 2 L \begin{aligned} a_{com} &= \frac{d^2 }{d t^2} \frac{q^2 t^2 }{\rho^2 A^2 L}\\ &=\frac{2 q^2}{\rho^2 A^2 L} \end{aligned}

Finally, we can substitute this acceleration into F = m a F = ma to find the corresponding force.

F f l o w = m s a n d 2 q 2 ρ 2 A 2 L = L A ρ 2 q 2 ρ 2 A 2 L = 2 q 2 ρ A \begin{aligned} F_{flow} &= m_{sand} \frac{2 q^2}{\rho^2 A^2 L}\\ &= L A \rho \frac{2 q^2}{\rho^2 A^2 L}\\ &= \frac{2 q^2}{\rho A} \end{aligned}

Thus the total force on the scale is the force of gravity, plus this force of the moving center of mass due to flowing sand:

F = M g + 2 q 2 ρ A F = Mg + \frac{2 q^2}{\rho A}

Boy I wish someone would do a test to put this to a rest.

Michael Mendrin - 3 years ago

Log in to reply

Your wish is my command, Michael. https://aapt.scitation.org/doi/full/10.1119/1.4973527

It has a video and everything. Enjoy!

Blake Farrow Staff - 3 years ago

Log in to reply

oh finally! this one has bugged me for years... thanks!

Michael Mendrin - 3 years ago

Yeah, it's all about the acceleration of the COM. The scale is completely indifferent to the internal forces involved.

Steven Chase - 3 years ago

Log in to reply

How could the scale be indifferent to the internal forces involved? What would push down on the scale if not the weight of the sand and it's impact on the bottom part of the hourglass?

I have made this discussion , in which I talk about this and some other doubts relating this problem, I'd be really happy if you shared your thougts there

Pedro Cardoso - 3 years ago

This is very similar to an Indian national physics Olympiad problem(2018) Good work on the problem tho Liked it

Suhas Sheikh - 3 years ago

Maybe for a follow-up problem, we can make the upper and lower halves of the hourglass have different geometries.

Steven Chase - 3 years ago

Please allow me ask a simple question, as I can't get my head around it: The center of mass is moving downwards with the sand going from top to bottom. With time going by, less sand is in the upper half, thus the pressure there decreases. And so the speed of the COM movement decreases with time aswell. This means that the acceleration is pointed upwards, so the hourglass must be lighter when in flow, not heavier. However, all the given answers make it heaver while in flow.

What did i get wrong?

Sven Roemer - 3 years ago

Log in to reply

If acceleration is upwards, the weight increases. Rockets accelerate upwards, so do astronauts inside, and they experience overloads.

Pavlo Bulanchuk - 3 years ago

You make a really solid point Sven, it made me take another look at my solution and Arjen's.

In the system's natural units, here is the movement of the sand's center of mass. I've made red lines showing the boundaries between the upper and lower halves, and for simplicity reduced the length of the intermediate path to zero -- this doesn't change much, though I'm happy to entertain more detailed models. The center of mass moves from it's initial position halfway up the top part of the hourglass, to the final position halfway up the bottom part.

The center of mass is moving downwards, and the rate of this motion is maximum when the sand just starts flowing: this initial acceleration is nearly instantaneous and the center of mass begins moving downwards. It proceeds to get slower as the position of the bottom sand's center of mass moves up (as that part of the hourglass fills). Just as you said, the acceleration is pointed upwards, so it's not much of an acceleration at all, but a deceleration.

Well, what is causing the deceleration of the sand's center of mass? A decelerating mass must be decelerated by a force, and that upwards force comes from the scale. It's as if in very slow motion, the mass of sand is moving down, but its motion is being halted as it runs up against the hourglass bottom and the scale.

The weight that a scale measures is the upwards normal force (or reaction force) that the scale exerts on whatever is on it. As Newton's laws state, this normal force is equal and opposite to the weight of the object on the scale.

I just thought through this while walking my dog, so let me know if it's at all unclear. This is certainly the problem that keeps on giving!

Blake Farrow Staff - 3 years ago

That was a great problem from which I learned a lot, thanks! Couldn't we just use Bernoulli's equation and consider the pressure due to velocity, which is rho·v²/2? Then, by using that pressure=force/A we get F=rho·v²·A/2. On the other hand we know that F=dp/dt=m·a+v·dm/dt, where q=dm/dt, so the excess of weight is just F=vq. We join both F=F and get q=rho·v·A/2 and then v=2·q/(rho·A). We plug this into v·q to get F=2q²/(rho·A)

al al - 3 years ago
Arsenio Meneses
May 25, 2018

Arsenio Meneses - May 25/2018 ........ Lets consider that the top and bottom part have area A and height L and are connected directly. So, the center of Mass of the system will move a distance L/2 + L/2 = L The time to follow will be M/q . So the acceleration of the C.M. will be given by L=(1/2)at^2 => a=2Lq^2/(M^2) The force acting on the C.M. while falling is Ma ===> F= 2Lq^2/M ===>> F= 2Lq^2/(δ.xVol) ==> F= 2Lq^2/δ.AL ==>. F= 2q^2/(δ.x A) δ - density The reading of the scale will be Mg + that F , or Mg+ 2q^2/(δ.xA), as the second option

Prince Zarzees
May 23, 2018

(h-y)=(1/2)g( t^2)----1

Mass flow rate is q,so average velocity

v=(q)/(density *A)

t=(h-y)/( (q)/(density*A ) Substituting t in 1

We get,

(h-y)g=2 (q^2)/((density A)^2)-----2 We know extra force=

A(h-y)(density*g)

Multiply both sides in 2 with (A*density)

So extra force=2(q^2)/(density*A)

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...