Skip to content
LumenaryAcademy
Mechanics 75 min

Dynamics of particles

Describing motion in three kinds of coordinates, Newton's second law as an equation you can integrate, energy and momentum methods, a projectile with air drag, and what happens in a collision.

Code tested · 5AI-draftedAwaiting review by a mechanical engineer

Your guides: Chloe · The Designer, Margaret · The Test Engineer, Gideon · The Checker

MotivationH

From balance to motion

In Statics and equilibrium every body stayed put, and Newton's second law collapsed to ∑F=0\sum \mathbf F = \mathbf 0. Now let the acceleration back in. A lift in a tower in Seoul accelerates its passengers upwards; a lorry brakes and the crate on its bed wants to keep going; a footballer in Lagos curls a free kick that the air slows and bends; two railway wagons meet in a shunting yard. In each case an engineer needs numbers. How hard does the cable pull? Will the crate slide? How far does the ball really go? How much energy must the buffers absorb?

This lesson treats bodies as particles: points with mass, whose size and rotation do not matter for the question asked. That is a judgement, not a fact about the body. A lorry is a particle when you ask how long it takes to stop, and very much not a particle when you ask whether it tips over on a bend; that second question belongs to Rigid-body dynamics. You met Newton's laws and stepped a model cart forward in time in Forces and motion, and energy stores in Energy and efficiency. Here those ideas become the tools of a first-year engineer: vectors and derivatives (vectors and calculus), differential equations solved step by step, and two conservation principles that often let you skip the solving altogether.

DefinitionsH

Kinematics: describing motion

Kinematics describes motion without asking what causes it. Its three quantities are linked by derivatives.

Definition— position, velocity, acceleration

A particle's position is a vector r(t)\mathbf r(t) from a fixed origin. Its velocity is v=dr/dt\mathbf v = \mathrm d\mathbf r/\mathrm dt and its acceleration is a=dv/dt=d2r/dt2\mathbf a = \mathrm d\mathbf v/\mathrm dt = \mathrm d^2\mathbf r/\mathrm dt^2. Speed is the magnitude v=∣v∣v = |\mathbf v|.

Rectilinear motion

Along a straight line everything is a scalar, and the useful question is what the acceleration depends on. If aa is given as a function of time (an accelerometer record, a motor's programmed profile), integrate twice. If it depends on speed (air drag) or on position (a spring), the chain rule gives a second form that is often easier:

a=dvdt=dvdxdxdt=v dvdx.a = \frac{\mathrm dv}{\mathrm dt} = \frac{\mathrm dv}{\mathrm dx}\frac{\mathrm dx}{\mathrm dt} = v\,\frac{\mathrm dv}{\mathrm dx}.

For constant aa this gives the familiar v2=v02+2a(x−x0)v^2 = v_0^2 + 2a(x - x_0); you will see in a moment that it is the work–energy principle in disguise.

Curvilinear motion: three coordinate systems

In the plane, r=x i+y j\mathbf r = x\,\mathbf i + y\,\mathbf j in Cartesian coordinates, and a=x¨ i+y¨ j\mathbf a = \ddot x\,\mathbf i + \ddot y\,\mathbf j: the components separate, which is why projectiles are usually done this way. (A dot means d/dt\mathrm d/\mathrm dt.) Two other systems use unit vectors that move with the particle, and they repay the extra effort when the geometry is curved.

PCρpath coordinatesetenOPrθpolar coordinatesereθ
Left: path (normal–tangential) coordinates. The unit vector e_t points along the path in the direction of motion and e_n points towards the centre of curvature C, at the radius of curvature ρ. Right: polar coordinates. The unit vector e_r points away from the origin O along r, and e_θ is at right angles to it, in the direction of increasing θ.

Definition— path (normal–tangential) coordinates

At each point of the path, et\mathbf e_t is the unit tangent in the direction of motion and en\mathbf e_n the unit normal towards the centre of curvature, at radius of curvature ρ\rho. Then v=v et\mathbf v = v\,\mathbf e_t and

a=v˙ et+v2ρ en.\mathbf a = \dot v\,\mathbf e_t + \frac{v^2}{\rho}\,\mathbf e_n .

The tangential part changes the speed; the normal part, the centripetal acceleration, changes the direction.

Definition— polar coordinates

With r=r er\mathbf r = r\,\mathbf e_r, where er=(cos⁡θ,sin⁡θ)\mathbf e_r = (\cos\theta, \sin\theta) and eθ=(−sin⁡θ,cos⁡θ)\mathbf e_\theta = (-\sin\theta, \cos\theta),

v=r˙ er+rθ˙ eθ,a=(r¨−rθ˙2) er+(rθ¨+2r˙θ˙) eθ.\mathbf v = \dot r\,\mathbf e_r + r\dot\theta\,\mathbf e_\theta, \qquad \mathbf a = (\ddot r - r\dot\theta^2)\,\mathbf e_r + (r\ddot\theta + 2\dot r\dot\theta)\,\mathbf e_\theta .

Polar coordinates suit anything measured from a pivot or a sensor: a radar tracking an aircraft by range and bearing, a slider on a rotating arm, a satellite. Their acceleration formula looks arbitrary until you see where its four terms come from.

Theorem— Acceleration in polar coordinates

If r(t)r(t) and θ(t)\theta(t) are twice differentiable, the acceleration of r=r er\mathbf r = r\,\mathbf e_r is (r¨−rθ˙2) er+(rθ¨+2r˙θ˙) eθ(\ddot r - r\dot\theta^2)\,\mathbf e_r + (r\ddot\theta + 2\dot r\dot\theta)\,\mathbf e_\theta.

Proof

The unit vectors turn with θ\theta: differentiating their components gives e˙r=θ˙ eθ\dot{\mathbf e}_r = \dot\theta\,\mathbf e_\theta and e˙θ=−θ˙ er\dot{\mathbf e}_\theta = -\dot\theta\,\mathbf e_r. By the product rule, v=r˙ er+r e˙r=r˙ er+rθ˙ eθ\mathbf v = \dot r\,\mathbf e_r + r\,\dot{\mathbf e}_r = \dot r\,\mathbf e_r + r\dot\theta\,\mathbf e_\theta. Differentiating again, a=r¨ er+r˙θ˙ eθ+r˙θ˙ eθ+rθ¨ eθ+rθ˙ (−θ˙ er)\mathbf a = \ddot r\,\mathbf e_r + \dot r\dot\theta\,\mathbf e_\theta + \dot r\dot\theta\,\mathbf e_\theta + r\ddot\theta\,\mathbf e_\theta + r\dot\theta\,(-\dot\theta\,\mathbf e_r), and collecting terms gives the result.

∎

Example— A gripper sliding along a turning arm

A pick-and-place robot's arm turns at a steady θ˙=2.0\dot\theta = 2.0 rad/s while its gripper slides outwards along the arm at a steady r˙=0.50\dot r = 0.50 m/s. At r=0.80r = 0.80 m, with r¨=0\ddot r = 0 and θ¨=0\ddot\theta = 0:

ar=−rθ˙2=−0.80×2.02=−3.2 m/s2,aθ=2r˙θ˙=2×0.50×2.0=2.0 m/s2.a_r = -r\dot\theta^2 = -0.80 \times 2.0^2 = -3.2 \text{ m/s}^2, \qquad a_\theta = 2\dot r\dot\theta = 2 \times 0.50 \times 2.0 = 2.0 \text{ m/s}^2 .

The magnitude is 3.22+2.02=3.8\sqrt{3.2^2 + 2.0^2} = 3.8 m/s². A 2.0 kg part held in the gripper therefore needs a force of about 7.5 N from it, pointing inwards and sideways, even though neither rr nor θ\theta is accelerating. If the gripper's hold is rated only for the part's weight with a margin, that sideways push matters.

ComputationC

Rectilinear motion from an accelerometer record

A lift moves in a straight line, and its passengers are its sensors. Its drive never switches acceleration on and off abruptly: the rate of change of acceleration, the jerk, is felt as a jolt, so controllers ramp the acceleration smoothly. The program below builds an acceleration record like the one an accelerometer on the cabin floor would log, integrates it twice with the trapezoidal rule (a numerical method you can trust when the samples are close together), and applies Newton's second law to a passenger standing on a bathroom scale.

python
import numpy as np

# A passenger lift's acceleration, as an accelerometer would record it (upwards positive),
# sampled 20 times a second. Each speed change is a smooth pulse a_max * sin^2(pi t / T).
a_max, T = 1.2, 6.0           # peak acceleration (m/s^2) and pulse length (s)
cruise = 26.0                 # seconds at steady speed between the two pulses
dt = 0.05
t = np.arange(0.0, 2 * T + cruise + dt / 2, dt)
pulse = lambda s: np.where((s >= 0) & (s <= T), a_max * np.sin(np.pi * s / T) ** 2, 0.0)
a = pulse(t) - pulse(t - T - cruise)

# Integrate twice with the trapezoidal rule: v = integral of a, x = integral of v.
v = np.concatenate([[0.0], np.cumsum((a[1:] + a[:-1]) / 2 * dt)])
x = np.concatenate([[0.0], np.cumsum((v[1:] + v[:-1]) / 2 * dt)])
jerk = np.gradient(a, dt)     # rate of change of acceleration, felt as a jolt

print(f"top speed {v.max():.2f} m/s, travel {x[-1]:.1f} m in {t[-1]:.0f} s, "
      f"final speed {abs(v[-1]):.3f} m/s, peak jerk {abs(jerk).max():.2f} m/s^3")

# Newton's second law for a 70 kg passenger standing on a scale: N - m g = m a.
m, g = 70.0, 9.81
N = m * (g + a)
print(f"scale reading: {N.min() / g:.1f} kg to {N.max() / g:.1f} kg (at rest: {m:.0f} kg)")

print("\n  t (s)   a (m/s2)  v (m/s)   x (m)   scale (kg)")
for ti in [0, 3, 6, 10, 20, 32, 35, 38]:
    i = int(round(ti / dt))
    print(f"{t[i]:6.1f} {np.round(a[i], 2) + 0.0:9.2f} {np.round(v[i], 2) + 0.0:8.2f} "
          f"{x[i]:8.1f} {N[i] / g:10.1f}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

The lift climbs 115.2 m, about thirty storeys, in 38 s, cruising at 3.60 m/s. Each acceleration pulse adds amax⁡T/2=3.6a_{\max}T/2 = 3.6 m/s of speed, because sin⁡2\sin^2 averages to one half, and covers amax⁡T2/4=10.8a_{\max}T^2/4 = 10.8 m; the cruise covers 3.6×26=93.63.6 \times 26 = 93.6 m. The free-body diagram of the passenger has two forces, the weight mgmg down and the scale's push NN up, so N−mg=maN - mg = ma and the scale reads m(g+a)/gm(g + a)/g: 78.6 kg at the peak of the upward acceleration and 61.4 kg while slowing down at the top.

DefinitionsH

Newton's second law, in any coordinates

Definition— Newton's second law for a particle

In an inertial frame of reference, a particle of constant mass mm moves so that

∑F=m a,\sum \mathbf F = m\,\mathbf a ,

where the sum runs over every force acting on it. Written in components, it is two equations in the plane (three in space) in whichever coordinates suit the problem: ∑Fx=mx¨\sum F_x = m\ddot x; or ∑Ft=mv˙\sum F_t = m\dot v and ∑Fn=mv2/ρ\sum F_n = mv^2/\rho; or ∑Fr=m(r¨−rθ˙2)\sum F_r = m(\ddot r - r\dot\theta^2) and ∑Fθ=m(rθ¨+2r˙θ˙)\sum F_\theta = m(r\ddot\theta + 2\dot r\dot\theta).

The working method extends the one from statics. Draw the free-body diagram, with every force from everything you cut away; next to it draw the kinetic diagram, a single arrow mam\mathbf a; and equate the two, component by component. Problems come in two directions. Either the motion is known and you want forces (the gripper above, the lift's rope), which is algebra; or the forces are known and you want the motion, which means solving a differential equation, by hand when it is simple and numerically when it is not.

Example— Will the crate slide when the lorry brakes?

An unsecured wooden crate sits on the steel bed of a lorry, with a static friction coefficient of μs=0.35\mu_s = 0.35 between them. The lorry brakes at 4.04.0 m/s². In the ground's frame, the only horizontal force on the crate is friction from the bed, so to decelerate with the lorry it needs F=m×4.0F = m \times 4.0. The bed can supply at most μsN=μsmg=m×0.35×9.81=m×3.43\mu_s N = \mu_s mg = m \times 0.35 \times 9.81 = m \times 3.43. The mass cancels, and 3.43<4.03.43 \lt 4.0: friction is not enough, and the crate slides forwards along the bed towards the cab.

This is why load-restraint rules exist. An emergency stop can be harder still, more than friction alone will hold for almost any load, so loads are lashed, blocked against the headboard or packed tightly, to restraint capacities calculated by load-securing standards such as Europe's EN 12195-1 and the guidance of national road authorities.

Example— A car on a flat roundabout

A car goes round a flat roundabout of radius 50 m at a steady 50 km/h (13.9 m/s). In path coordinates at=0a_t = 0 and an=v2/ρ=13.92/50=3.86a_n = v^2/\rho = 13.9^2/50 = 3.86 m/s², towards the centre, and only sideways friction at the tyres can supply it: the road must provide μ≥an/g=0.39\mu \ge a_n/g = 0.39, easy on dry asphalt and impossible on ice. Railways and fast roads tilt the surface (cant or superelevation) so that the normal force supplies part of the centripetal force.

DefinitionsH

Work–energy and impulse–momentum

Newton's second law is a differential equation. Integrating it once, either along the path or over time, gives two principles that often answer a question without solving for the whole motion.

Definition— work, kinetic energy, power

The work done by a force F\mathbf F as its point of application moves from position 1 to position 2 is U1→2=∫12F⋅drU_{1\to 2} = \int_1^2 \mathbf F \cdot \mathrm d\mathbf r. A particle's kinetic energy is T=12mv2T = \tfrac12 m v^2. Power is the rate of doing work, P=F⋅vP = \mathbf F \cdot \mathbf v, in watts.

Theorem— Work–energy principle

For a particle of constant mass, the total work done by all the forces on it equals its change in kinetic energy: ∑U1→2=T2−T1\sum U_{1\to2} = T_2 - T_1.

Proof

ddt(12m v⋅v)=m a⋅v=(∑F)⋅v\dfrac{\mathrm d}{\mathrm dt}\left(\tfrac12 m\,\mathbf v\cdot\mathbf v\right) = m\,\mathbf a\cdot\mathbf v = \left(\sum\mathbf F\right)\cdot\mathbf v. Integrate both sides over time from state 1 to state 2 and use v dt=dr\mathbf v\,\mathrm dt = \mathrm d\mathbf r.

∎

Some forces do work that depends only on the end points: gravity, −mg Δh-mg\,\Delta h, and a linear spring, −12k(x22−x12)-\tfrac12 k(x_2^2 - x_1^2). These are conservative, and their work can be booked as changes in potential energy; friction and drag are not, and turn kinetic energy into heat.

Example— Sizing an end-stop buffer

An overhead crane's trolley, 4.0 t with its hook block, runs along its rails at 1.2 m/s. If its brakes fail, buffers at the end of the rails must stop it within 0.10 m. With a linear spring buffer, the work–energy principle gives 12kx2=12mv2\tfrac12 k x^2 = \tfrac12 m v^2, so k=mv2/x2=4000×1.22/0.102=576k = mv^2/x^2 = 4000 \times 1.2^2/0.10^2 = 576 kN/m and the peak force is kx=57.6kx = 57.6 kN, a peak deceleration of 14.4 m/s², about 1.5 gg.

A hydraulic buffer is designed to push back with a nearly constant force instead. The same energy, 12mv2=2.88\tfrac12 mv^2 = 2.88 kJ, absorbed over the same 0.10 m needs only F=2.88/0.10=28.8F = 2.88/0.10 = 28.8 kN: half the peak force, from simply shaping the force–displacement curve. That is why buffers on cranes, lift pits and rail end-stops are so often hydraulic or made of energy-absorbing elastomers, and why the rules for lifts and cranes specify buffers at all.

Theorem— Impulse–momentum principle

The linear momentum of a particle is p=mv\mathbf p = m\mathbf v, and ∑F=dp/dt\sum\mathbf F = \mathrm d\mathbf p/\mathrm dt. Integrating over time, the impulse of the forces equals the change in momentum: ∫t1t2∑F dt=mv2−mv1\int_{t_1}^{t_2} \sum\mathbf F\,\mathrm dt = m\mathbf v_2 - m\mathbf v_1. For a system of particles the internal forces cancel in pairs (Newton's third law), so if the external impulse is zero, total momentum is conserved.

The impulse needed to stop something is fixed by its momentum; the force is not. A 0.16 kg cricket ball arriving at 30 m/s carries 4.8 N s of momentum. Stopped by stiff hands in 0.01 s, the average force is 480 N; a fielder who lets the hands ride back with the ball over 0.1 s feels 48 N. Crumple zones, airbags, cycle helmets and the soft landing mats of a gym all work the same way: they lengthen the time, or the distance, over which a fixed amount of momentum or energy is removed.

ComputationC

A projectile with air drag

In a vacuum, a projectile launched from level ground at speed v0v_0 and angle θ\theta has x¨=0\ddot x = 0 and y¨=−g\ddot y = -g, so x=v0tcos⁡θx = v_0 t\cos\theta, y=v0tsin⁡θ−12gt2y = v_0 t\sin\theta - \tfrac12 g t^2, a parabola with range R=v02sin⁡2θ/gR = v_0^2\sin 2\theta/g, greatest at θ=45°\theta = 45°. Real air pushes back. For a ball moving fast enough to leave a turbulent wake (almost every ball in sport), the drag force is well described by

FD=−12ρ CdA ∣v∣ v,\mathbf F_D = -\tfrac12 \rho\, C_d A\, |\mathbf v|\,\mathbf v ,

where ρ≈1.2\rho \approx 1.2 kg/m³ is the density of air, AA the ball's frontal area and CdC_d its drag coefficient, about 0.4–0.5 for a sphere over a wide range of speeds. Newton's second law then reads, with k=ρCdA/(2m)k = \rho C_d A / (2m),

x¨=−k ∣v∣ x˙,y¨=−g−k ∣v∣ y˙.\ddot x = -k\,|\mathbf v|\,\dot x, \qquad \ddot y = -g - k\,|\mathbf v|\,\dot y .

The two directions are now coupled through ∣v∣|\mathbf v|, and there is no closed-form solution. When drag balances weight in a vertical fall, the ball has reached its terminal speed vt=g/k=2mg/(ρCdA)v_t = \sqrt{g/k} = \sqrt{2mg/(\rho C_d A)}, and the ratio v0/vtv_0/v_t tells you at a glance whether drag will matter.

So integrate numerically, as an ordinary differential equation for the state (x,y,x˙,y˙)(x, y, \dot x, \dot y). The program uses the classical fourth-order Runge–Kutta method, which samples the slope four times per step and has an error that shrinks like the fourth power of the step; it launches each ball at 141 angles at once, as NumPy arrays, and interpolates the moment each one crosses the ground.

python
import numpy as np

g, rho = 9.81, 1.20                 # gravity (m/s^2) and density of air (kg/m^3)

def flights(m, d, cd, v0, angles_deg, dt=0.002):
    """Launch one ball at many angles at once, from ground level, with quadratic drag.
    Returns the range and the highest point for each angle (metres)."""
    k = rho * cd * (np.pi * d**2 / 4) / (2 * m)        # drag force / mass = k |v| v
    th = np.radians(angles_deg)
    s = np.array([0 * th, 0 * th, v0 * np.cos(th), v0 * np.sin(th)])   # rows: x, y, vx, vy

    def rates(s):                                      # ds/dt from Newton's second law
        x, y, vx, vy = s
        speed = np.hypot(vx, vy)
        return np.array([vx, vy, -k * speed * vx, -g - k * speed * vy])

    reach, top = np.full(th.shape, np.nan), np.zeros(th.shape)
    while np.isnan(reach).any():                       # classical fourth-order Runge-Kutta
        k1 = rates(s)
        k2 = rates(s + dt / 2 * k1)
        k3 = rates(s + dt / 2 * k2)
        k4 = rates(s + dt * k3)
        new = s + dt / 6 * (k1 + 2 * k2 + 2 * k3 + k4)
        top = np.maximum(top, new[1])
        hit = np.isnan(reach) & (new[1] < 0)           # crossed the ground in this step:
        f = s[1, hit] / (s[1, hit] - new[1, hit])      # interpolate where
        reach[hit] = s[0, hit] + f * (new[0, hit] - s[0, hit])
        s = new
    return reach, top

angles = np.arange(10.0, 80.01, 0.5)
balls = [  # name, mass (kg), diameter (m), drag coefficient, launch speed (m/s)
    ("shot put", 7.26, 0.120, 0.47, 13.5),
    ("football", 0.43, 0.220, 0.25, 25.0),
    ("cricket ball", 0.16, 0.072, 0.50, 30.0),
    ("shuttlecock", 0.0050, 0.065, 0.60, 50.0),
]
print("               v0    v_t   range at 45 deg       best angle with drag")
print("             (m/s)  (m/s)  vacuum  with drag     angle   range")
for name, m, d, cd, v0 in balls:
    vt = np.sqrt(2 * m * g / (rho * cd * np.pi * d**2 / 4))    # terminal speed
    reach, top = flights(m, d, cd, v0, angles)
    i45, best = np.argmin(abs(angles - 45)), np.argmax(reach)
    vac = v0**2 / g                                         # vacuum range at 45 degrees
    print(f"{name:12s} {v0:5.1f} {vt:6.1f}  {vac:6.1f}  {reach[i45]:6.1f}      "
          f"{angles[best]:4.1f}  {reach[best]:6.1f}")

# Check the integrator against the exact vacuum result (drag coefficient zero).
r0, h0 = flights(0.43, 0.22, 0.0, 25.0, np.array([30.0, 45.0, 60.0]))
exact = 25.0**2 * np.sin(np.radians([60.0, 90.0, 120.0])) / g
print("\nvacuum check, max range error:", f"{abs(r0 - exact).max():.1e} m")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

The last line is the habit worth copying: before trusting a simulation of something you cannot solve, run it on something you can. With Cd=0C_d = 0 it reproduces the parabola to a few micrometres.

The table sorts the four balls by v0/vtv_0/v_t. A shot put (launched here from the ground, ignoring the athlete's two-metre release height) has a terminal speed of 149 m/s, ten times its launch speed: drag costs it 0.1 m in 18.6 and the vacuum theory is fine. A football kicked at 25 m/s is launched at about its terminal speed, and drag cuts its range from 63.7 m to 39.9 m, more than a third. A badminton shuttlecock hit at 50 m/s is at eight times its terminal speed of 6.4 m/s. The vacuum formula would send it 255 m; the air stops it at about 11 m, and its best angle is no longer 45° but about 27°.

0102030405060 min a vacuum: 63.7 m, 15.9 m highwith drag: 39.9 m, 12.1 m high
The football from the table, kicked at 25 m/s and 45°, drawn to scale from the same integration. Dashed: in a vacuum, a symmetric parabola 63.7 m long and 15.9 m high. Solid: with drag, 39.9 m long and 12.1 m high, and steeper coming down than going up.

Note

The weakest number in this program is CdC_d, not the integrator. For a smooth sphere the drag coefficient drops sharply, from about 0.5 to below 0.2, when the boundary layer turns turbulent (the drag crisis), and a ball's seams, panels, dimples and spin move where that happens. Footballs, cricket balls and golf balls are engineered around it, which is why a ball's range is measured in wind tunnels and on the pitch, not taken from a table.

Worked examplesH

Collisions and restitution

When two bodies collide, they push on each other with a large force for a short time: well under a millisecond for two steel balls, about a hundredth of a second for a kicked football, a fraction of a second for railway wagons with sprung buffers. Over so short a time the ordinary forces (weights, friction, drag) deliver negligible impulse, so the pair's total momentum is conserved through the impact. Momentum alone gives one equation for two unknown final velocities; the second comes from how springy the contact is.

Definition— line of impact, coefficient of restitution

The line of impact is the common normal to the surfaces at the point of contact. An impact is central if the centres of mass of both bodies lie on it. For a central impact in which the bodies approach with velocities u1u_1, u2u_2 and separate with v1v_1, v2v_2 (components along the line of impact), the coefficient of restitution is

e=v2−v1u1−u2,0≤e≤1,e = \frac{v_2 - v_1}{u_1 - u_2}, \qquad 0 \le e \le 1,

the speed of separation divided by the speed of approach. e=1e = 1 is a perfectly elastic impact and e=0e = 0 a perfectly plastic one, after which the bodies move together.

Momentum, m1u1+m2u2=m1v1+m2v2m_1u_1 + m_2u_2 = m_1v_1 + m_2v_2, and restitution together give

v1=m1u1+m2u2−m2e(u1−u2)m1+m2,v2=m1u1+m2u2+m1e(u1−u2)m1+m2.v_1 = \frac{m_1u_1 + m_2u_2 - m_2 e(u_1 - u_2)}{m_1 + m_2}, \qquad v_2 = \frac{m_1u_1 + m_2u_2 + m_1 e(u_1 - u_2)}{m_1 + m_2}.

Theorem— Kinetic energy lost in a central impact

In a central impact, the kinetic energy dissipated is ΔT=12 μ (1−e2) (u1−u2)2\Delta T = \tfrac12\,\mu\,(1 - e^2)\,(u_1 - u_2)^2, where μ=m1m2/(m1+m2)\mu = m_1m_2/(m_1 + m_2) is the reduced mass.

Proof

Write each velocity as the velocity of the centre of mass, V=(m1u1+m2u2)/(m1+m2)V = (m_1u_1 + m_2u_2)/(m_1 + m_2), plus a velocity relative to it. The kinetic energy splits into 12(m1+m2)V2\tfrac12(m_1 + m_2)V^2, which is unchanged because momentum is conserved, plus the energy of the relative motion, which a short calculation shows is 12μw2\tfrac12\mu w^2 for relative velocity w=u1−u2w = u_1 - u_2. Restitution multiplies ww by −e-e, so the relative energy falls from 12μw2\tfrac12\mu w^2 to 12μe2w2\tfrac12\mu e^2w^2.

∎

beforeduringafter1: 30 t2: 20 t1212u₁ = 2.0 m/sat rest0.72 m/s1.92 m/sFF
A central impact of two railway wagons. Before: wagon 1 (30 t) runs at u₁ = 2.0 m/s towards wagon 2 (20 t) at rest. During the impact the buffers push on the two wagons with equal and opposite forces, so momentum is conserved. After, with e = 0.6: v₁ = 0.72 m/s and v₂ = 1.92 m/s.

The program below works through the figure's wagons for four values of ee and then shows how ee is measured. For a ball dropped onto a hard floor, the impact speed after a fall from rest through hh is 2gh\sqrt{2gh}, so the ratio of rebound to drop height is e2e^2. The drop test used to approve tennis balls works exactly this way: a ball released from rest at 254 cm onto concrete must rebound to roughly 135–147 cm.

python
import numpy as np

def collide(m1, u1, m2, u2, e):
    """Velocities after a central impact: momentum is conserved and the relative
    velocity of separation is e times the relative velocity of approach."""
    p = m1 * u1 + m2 * u2
    v1 = (p - m2 * e * (u1 - u2)) / (m1 + m2)
    v2 = (p + m1 * e * (u1 - u2)) / (m1 + m2)
    return v1, v2

# A 30 t railway wagon rolling at 2.0 m/s runs into a stationary 20 t wagon.
m1, u1, m2, u2 = 30e3, 2.0, 20e3, 0.0
T0 = 0.5 * m1 * u1**2
print("   e    v1 (m/s)  v2 (m/s)  momentum (kN s)  energy lost (kJ)")
for e in [1.0, 0.6, 0.3, 0.0]:
    v1, v2 = collide(m1, u1, m2, u2, e)
    lost = T0 - 0.5 * m1 * v1**2 - 0.5 * m2 * v2**2
    print(f"{e:4.1f}  {v1:9.2f} {v2:9.2f} {(m1 * v1 + m2 * v2) / 1e3:12.1f} {lost / 1e3:15.1f}")

mu = m1 * m2 / (m1 + m2)                   # the "reduced mass"
print(f"check, e = 0: (1/2) mu (1 - e^2) u_rel^2 = {0.5 * mu * u1**2 / 1e3:.1f} kJ")

# Measuring e: a ball released from rest 2.54 m above a hard floor rebounds to 1.41 m.
# Rebound speed / impact speed = sqrt(h1 / h0), because v = sqrt(2 g h) for each fall.
g, h0, h1 = 9.81, 2.54, 1.41
e = np.sqrt(h1 / h0)
n = np.arange(6)
heights = h0 * e ** (2 * n)                # each bounce reaches e^2 times the last height
t_first = np.sqrt(2 * h0 / g)              # time of the first fall
print(f"\ne = {e:.3f}; bounce heights (m):", np.round(heights, 2))
print(f"time until the bouncing stops, in this model: {t_first * (1 + e) / (1 - e):.2f} s")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Momentum is 60.0 kN s in every row; what changes is where the energy goes. When the wagons couple and move off together at 1.20 m/s (e=0e = 0), 24.0 kJ of the original 60 kJ, 40 per cent, is turned into heat, noise and permanent deformation in the buffers and couplings. With e=1e = 1 none is lost, and the struck wagon leaves at 2.40 m/s, faster than the one that hit it, because it is lighter.

For an oblique impact of smooth bodies, the impulse acts only along the line of impact. Resolve the velocities along and across that line: the components across it are unchanged, and the components along it obey the two equations above. One elegant consequence, familiar to anyone who plays snooker in Sheffield or carrom in Chennai: when one disc or ball strikes an identical one at rest, off-centre, with ee close to 1, the two move off at right angles to each other.

History & global context

From impetus to Runge–Kutta

Aristotle taught that a thrown stone keeps moving only while something keeps pushing it. Ibn Sīnā in eleventh-century Persia and Jean Buridan in fourteenth-century Paris replaced the push with an internal impetus given to the stone at launch, a step towards the idea of momentum. Galileo, rolling balls down inclined planes, showed that falling bodies gain speed uniformly and that a projectile's path is a parabola (Two New Sciences, 1638), noting that air resistance spoils it for light or fast bodies. In 1668–69 John Wallis, Christopher Wren and Christiaan Huygens answered a Royal Society challenge on collisions with the conservation of momentum, and Newton's Principia (1687) gave the laws of motion and an experimental law of restitution in the form used above. Leonhard Euler wrote the second law as differential equations in components in 1750, and Gaspard-Gustave de Coriolis analysed the extra terms of rotating frames in 1835, while studying waterwheels and machines. Francis Bashforth in England measured the drag on fast projectiles systematically from the 1860s, and around 1900 Carl Runge and Wilhelm Kutta devised the step-by-step integration methods every simulation here relies on.

ExplanationH

Particle dynamics in professional practice

Simulation at scale. The integrator in this lesson is a teaching version of what runs inside every dynamics package. In Python, engineers usually call SciPy's solve_ivp, which adapts its step size to meet an error tolerance; large codes use libraries such as LLNL's SUNDIALS. Multibody and particle engines, commercial and open-source (such as Project Chrono and MuJoCo), simulate thousands to millions of interacting bodies: grains in a hopper, ballast under a railway sleeper, robots, whose rotating links robot dynamics in Lumenary Robotics treats.

Impact and crash safety. Vehicles are designed to manage exactly the impulse and energy of this lesson's collisions: crumple zones lengthen the stopping distance, and restraints couple the occupant to the slowing car early. New-car assessment programmes test them, among them Euro NCAP, the programmes of NHTSA and IIHS in the United States, C-NCAP in China, Bharat NCAP in India, Latin NCAP and ASEAN NCAP. Before a car is crashed it has been crashed thousands of times in explicit finite-element codes, commercial and open-source (OpenRadioss, open since 2022).

Lifts, cranes and space. Lifts and cranes are designed to codes from bodies such as ASME (the A17.1 code for elevators and escalators, the B30 series for cranes) and CEN in Europe, which require the buffers, overspeed governors and safety gear whose sizing is particle dynamics. Spaceflight is the purest particle dynamics of all: a spacecraft between planets is a point mass in the gravity of the Sun and planets, integrated with great care by mission teams at agencies such as NASA, ESA, ISRO and JAXA.

What is safe to try. Time a toy car on a ramp with a phone camera, measure a ball's bounce against a tape measure on a wall, or film a gently thrown soft ball and compare its path with the simulation. Anything involving heavy or fast objects, loads on vehicles or hoists is engineered, inspected and operated by trained people, for the reasons this lesson computes.

Common misconceptionsH

Common misconceptions

Misconception

"A car going round a bend at a steady speed is not accelerating." Acceleration is the rate of change of the velocity vector. On a curve the direction changes, so there is a centripetal acceleration v2/ρv^2/\rho, and a force (here tyre friction) must supply it.

Misconception

"The angle for maximum range is always 45°." Only in a vacuum, on level ground. Drag lowers the best angle (to about 27° for the shuttlecock above), launching from a height lowers it further, and an athlete who can throw faster at a flatter angle chooses a lower one still. Shot-putters typically release at under 40°.

Misconception

"In a collision, energy is conserved, so the bodies bounce apart as fast as they came together." Total energy is conserved, but kinetic energy is not unless e=1e = 1. Momentum is the quantity that is conserved through any short impact between two bodies free of large outside forces.

Misconception

"All objects fall at the same rate, whatever their mass." Only in a vacuum, where every body falls with acceleration gg. In air it is only roughly true. For two balls of the same size and shape, the heavier has the higher terminal speed 2mg/(ρCdA)\sqrt{2mg/(\rho C_d A)} and does fall faster, which is why a shot put ignores the air and a shuttlecock is ruled by it.

Exercises

Exercises

Use g=9.81g = 9.81 m/s² throughout. Two exercises ask for Python functions; run them in the editor before submitting.

1. Dynamics of particles

Warm-upHuman

A car drives round a roundabout at a steady 30 km/h. Which statement about its acceleration is true?

Choose one

2. Dynamics of particles

CoreHuman

A car rounds a flat (uncanted) curve of radius 40 m. The coefficient of static friction between its tyres and the road is 0.70. What is the highest steady speed at which it can take the curve without sliding outwards, in km/h, rounded to the nearest whole number?

3. Dynamics of particles

CoreHuman

A bead slides outwards along a straight rod at a steady speed r˙\dot r relative to the rod, while the rod turns about one end at a steady angular speed θ˙\dot\theta. Which terms of the polar-coordinate acceleration a=(r¨−rθ˙2) er+(rθ¨+2r˙θ˙) eθ\mathbf a = (\ddot r - r\dot\theta^2)\,\mathbf e_r + (r\ddot\theta + 2\dot r\dot\theta)\,\mathbf e_\theta are non-zero? Select all that apply.

Choose all that apply

4. Dynamics of particles

CoreHuman

A 2.0 t crane trolley moving at 1.5 m/s must be stopped by a linear spring buffer within 0.15 m of compression. What is the smallest spring stiffness that does this, in kN/m?

5. Dynamics of particles

CoreHuman

A 1200 kg car travelling at 15 m/s runs into the back of a stationary 1800 kg car, and the two lock together (e=0e = 0). How much kinetic energy is dissipated in the impact, in kJ?

6. Dynamics of particles

CoreComputational

Write collide(m1, u1, m2, u2, e) for a central impact of two bodies moving along one line, with velocities positive in the same direction. It returns (v1, v2, energy_lost): the velocities after the impact (m/s) and the kinetic energy dissipated (J). Use conservation of momentum and the definition of the coefficient of restitution.

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

7. Dynamics of particles

StretchComputational

Write flight_range(v0, angle_deg, m, d, cd, rho=1.2, g=9.81, dt=1e-3): the horizontal distance a ball travels when launched from ground level at speed v0 (m/s) and angle_deg above the horizontal, until it comes back down to ground level, with quadratic drag FD=−12ρCdA ∣v∣ v\mathbf F_D = -\tfrac12\rho C_d A\,|\mathbf v|\,\mathbf v, A=πd2/4A = \pi d^2/4.

Integrate the equations of motion step by step with time step dt (any sensible method: Runge–Kutta as in the lesson, or a small-step Euler method), and interpolate the landing point between the last two steps. The tests compare with the lesson's results to within about 1 per cent.

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Further reading and resources

  • MIT OpenCourseWare 2.003SC, Engineering Dynamics · Massachusetts Institute of TechnologyLectures, recitations and problem sets for a first university course in dynamics.
  • Mechanics Map · Jacob Moore and contributors, Penn StateAn open digital textbook whose dynamics chapters cover particle kinematics, kinetics, energy and impact with worked examples.
  • SciPy: scipy.integrate.solve_ivpThe adaptive Runge–Kutta and stiff solvers most Python engineers reach for when they integrate equations of motion.
  • NASA Glenn Research Center, Beginner's Guide to AeronauticsClear pages on the drag equation, drag coefficients and flight with and without air resistance.
  • Euro NCAPEurope's new-car assessment programme; its published test protocols show how impact dynamics is turned into crash tests.
  • Project ChronoAn open-source multibody and particle dynamics engine, used in vehicle, granular-flow and robotics research.