Circular Motion & Banked Turns · Guide · glidepathscience.org
GuideCircular Motion & Banked Turns
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook would take them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about fifteen minutes to read. Every section ends with a worked example that is not one of the sim's own problems. Everything here is described from the ground, where Newton's laws hold as written.
Part 1 · The essentials
Turning is accelerating
A car on a roundabout, a ball whirled on a string and the Moon in its orbit all move at a nearly steady speed, and yet each of them is accelerating. Acceleration is the rate of change of velocity, and velocity is a vector, with a direction as well as a size. A body moving on a circle keeps changing direction, so its velocity changes even when its speed does not. This matters in every problem here, because Newton's second law ties the net force to the acceleration, not to the speed.
Over a short time the velocity arrow turns a little toward the inside of the circle, so the change in velocity, the acceleration, points toward the center. This is the centripetal acceleration ("center-seeking"). Its size depends on the speed v and the radius r of the circle:
$a = v^2/r = \omega^2 r$
Here ω is the angular speed in radians per second, so that $v = \omega r$. Doubling the speed quadruples the acceleration: the velocity turns twice as fast and is twice as long. When the motion repeats every T seconds, the period, the body covers one circumference in that time:
$v = 2\pi r/T$ $\omega = 2\pi/T$
The most common mistake is to say that a body moving at constant speed has no acceleration, which is true only in a straight line. Another is to forget the square; the units catch it, since v/r is a rate of turning, in s⁻¹.
If every circling body accelerates toward the center, what happens when nothing makes it?
Nothing pulls outward: cut the string
Newton's first law says that a body with no net force on it keeps its velocity: the same speed in the same direction. A puck circling on a string has exactly one horizontal force on it, the string's pull toward the center. Cut the string and that force vanishes. From that instant the velocity stops changing, and the puck moves off in a straight line along the tangent, the direction it was moving when the string let go.
This settles a belief almost everyone brings to the subject: that something throws a circling body outward. If an outward force acted, a released puck would fly straight out along the radius. It leaves along the tangent instead, and its distance from the center grows only because a straight line drifts away from any circle it touches. The feeling of being pressed outward in a turning car is the same fact read the other way: your body keeps going straight until the door reaches you and pushes you inward. In the ground frame, which this sim uses throughout, there is no outward force to draw. (A turning observer's "centrifugal" force belongs to another frame, and to Accelerating Reference Frames.)
Since the inward pull is what keeps a body on its circle, the next question is how big that pull has to be.
The net force points to the center, and something real supplies it
Newton's second law, $\Sigma\vec F = m\vec a$, applies to circular motion like any other. In uniform circular motion the acceleration points to the center with size $v^2/r$, so the net force must point to the center too:
$\Sigma F = mv^2/r$, toward the center
"Centripetal" names the direction of this net force. It is not a new kind of force and never goes on a force diagram as an extra arrow. Some real force with a real agent supplies it: the string's tension for a puck, static friction for a car on a flat road, partly the normal force on a banked road, a drum's wall for a rotor rider, the Earth's gravity for the Moon. Draw the force diagram first, every force labeled by type and agent, and only then set the sum equal to $mv^2/r$.
For a puck on a frictionless air table, the table's push and the weight cancel, and the string, the only horizontal force, is the whole net force:
$T = mv^2/r$
The usual mistake is to add the weight to the tension, or to draw both a tension and a separate "centripetal force". There is one inward force here, and it is the string.
A string can hold a ball in a horizontal circle only if something else holds the ball up. Whirl it overhead, and the string has to do both jobs.
The conical pendulum
Swing a ball on a string around your head and the string sweeps out a cone. Two forces act on the ball: its weight mg, straight down, and the tension T along the string, which makes an angle θ with the vertical. The ball moves on a horizontal circle, so it has no vertical acceleration, and its horizontal acceleration points to the center of the circle. Splitting the tension into its vertical and horizontal parts gives one equation for each direction:
$T\cos\theta = mg$ $T\sin\theta = mv^2/r$
The vertical part of the tension holds the ball up; the horizontal part turns it. Dividing the two equations gives $\tan\theta = v^2/(rg)$: a faster swing opens the cone. But $T\cos\theta$ must always equal mg, and a horizontal string has no vertical part at all, so however fast the ball goes, the string can never be exactly level; the tension just climbs without limit. Many students expect a fast enough spin, or a tension above mg, to make the string level; but the tension exceeds mg at every angle, since $T = mg/\cos\theta$.
The relation $\tan\theta = v^2/(rg)$ will return, almost unchanged, when the road itself is tilted. First, though, the flat road.
A flat curve: static friction turns the car
A car rounding a flat curve has three forces on it: its weight, the road's upward normal force N, which balances the weight, and friction from the road on the tires. Only friction is horizontal, so friction must supply the whole inward force. Which friction is it? The car moves, but its tires roll without sliding sideways across the road, so the contact patch does not slip. The friction is static, and it points toward the center of the curve, where the acceleration points. Static friction supplies whatever the turn needs, up to a maximum of $\mu_s N = \mu_s mg$ on a level road. The fastest speed that holds is the one where the need reaches that maximum:
$\mu_s mg = mv^2/r$ $v_{\mathrm{max}} = \sqrt{\mu_s gr}$
The mass has canceled. A loaded truck gets more friction, but it needs more force to turn in exactly the same proportion, so its limit is the car's (ignoring real tires' differences). A wet road lowers $\mu_s$, and with it the safe speed, by the square root of the change.
Two mistakes are common here. One is to use the kinetic coefficient "because the car is moving"; kinetic friction appears only once the tires skid, and by then the car is already sliding wide. The other is to draw friction pointing outward, against an imagined outward push. Friction is the force that turns the car, so it points in.
A flat curve leaves friction to do all the work. Tilting the road lets the road's push help.
Banking the road
A road banked at an angle θ, with its outer edge higher, pushes on the car perpendicular to its surface. That normal force now tilts toward the center of the curve, so part of it can turn the car. Suppose the road is glare ice, with no friction at all. The only forces are the weight and the normal force, and the car moves on a horizontal circle, so the vertical parts must balance and the horizontal part must supply $mv^2/r$:
$N\sin\theta = mv^2/r$ $N\cos\theta = mg$
Dividing the first by the second gives the speed at which the bank alone turns the car, the design speed:
$\tan\theta = v^2/(rg)$ $v = \sqrt{rg\tan\theta}$
This is the conical pendulum's equation again, with the road's push in place of the string's pull. Like the flat-road limit, it does not depend on the mass. At the design speed no friction is needed, so on ice a car at exactly this speed holds its lane; any faster or slower and it slides.
The second equation also says something students rarely expect: $N = mg/\cos\theta$, which is more than the weight. On a ramp, where the block accelerates along the slope, the normal force is $mg\cos\theta$, less than the weight. On a banked curve the acceleration is horizontal, and the normal force must hold up the whole weight with only its vertical part. Riders on a steep bank feel heavier, not lighter.
Real roads are not glare ice. With friction, a banked curve holds a whole range of speeds, not one.
The window of safe speeds
On a banked road with friction, the design speed is only the middle of a window of speeds that hold. Above the design speed the tilted normal force cannot supply enough inward force by itself, and static friction must add some, pointing down the bank. Below the design speed the normal force tips the car toward the center more than the turn needs, and friction must pull the other way, up the bank. Either way friction can give at most $\mu_s N$, so the window has two edges. It is convenient to describe the friction's limit as an angle, the friction angle φ with $\tan\varphi = \mu_s$. The fastest and slowest speeds that hold are then
$v_{\mathrm{max}}^2 = rg\tan(\theta + \varphi)$ $v_{\mathrm{min}}^2 = rg\tan(\theta - \varphi)$
The bank and the friction combine as angles. If θ is no more than φ, the lower edge is zero, and even a stopped car stays put. The upper edge exists only while θ + φ < 90°. The surprise for most people is that a car can be too slow for a curve: on an icy, steep bank a car crawling far below the design speed slides down the bank, because gravity's pull down the slope is more than the ice can hold.
Curves and banks keep the motion level. A circle that stands upright brings gravity into the turn itself.
Over the top of a vertical circle
Swing a ball on a string in a vertical circle, or drive over the crest of a hill, and gravity points toward the center at the top of the circle and away from it at the bottom. At the very top, both the string's tension and the weight pull down, toward the center, so together they supply the inward force:
$T + mg = mv^2/r$
Gravity provides mg of the needed $mv^2/r$, so the string pulls less at the top. At the bottom it must also hold up the weight, $T - mg = mv^2/r$, so the tension is largest there. The common mistake at the top is to add the weight instead of subtracting it.
A string can only pull, so T cannot be negative. The slowest pass over the top with the string taut is the one where T reaches zero and gravity alone supplies the inward force:
$T = 0$: $mg = mv^2/r$ $v = \sqrt{gr}$
Any slower, and the string goes slack before the top: the ball leaves the circle and falls inside it until the string jerks taut. Over a hill crest the road pushes outward instead, $N = m(g - v^2/r)$: passengers feel light, and above $\sqrt{gr}$ the car leaves the road. In a dip they feel heavy.
Part 2 returns to these ideas with the acceleration written as a vector, speeds that change, and the full condition for friction on a bank.
Part 2 · Going further
The acceleration as a vector, and speeding up on a curve
The centripetal acceleration follows directly from calculus. Put the origin at the center and let the body turn at a steady angular speed ω, so its position is $\vec r(t) = r(\cos\omega t, \sin\omega t)$. Differentiating once gives the velocity, perpendicular to the position, and again gives the acceleration:
$\vec v = r\omega(-\sin\omega t, \cos\omega t)$ $\vec a = -\omega^2\vec r$
The acceleration is antiparallel to the position from the center, with size $\omega^2 r = v^2/r$. Each derivative brings out one factor of ω, which is why the acceleration goes as the square of the speed. Constant speed means $\vec v\cdot\vec a = 0$, an acceleration with no part along the velocity; it does not mean zero acceleration.
When the speed also changes, the acceleration gains a second part along the path, the tangential acceleration $a_t = dv/dt$. The two parts are perpendicular, so they combine by Pythagoras:
$a_c = v^2/r$ ${|a|} = \sqrt{a_{t}^2 + (v^2/r)^2}$
Speeding up through a curve tips the acceleration forward of the center; slowing down tips it behind. Adding the two parts as plain numbers is the usual slip. On any curved path the part perpendicular to the velocity is $v^2/\rho$, with ρ the local radius of curvature.
Both parts of that acceleration need a force, and on a road both come from the same tires.
The friction circle: braking in a curve
On a flat road the tires' grip is a single budget, at most $\mu_s N = \mu_s mg$, and every horizontal force the car needs comes out of it. Turning needs $mv^2/r$ across the road; braking or speeding up needs $ma_t$ along it. The two are perpendicular, so the grip they draw on together is the length of their sum:
$\sqrt{a_{t}^2 + (v^2/r)^2} \le \mu_s g$
(A model: it assumes the tires grip equally in every direction and braking shifts no weight forward.) Drawn on axes of $a_t$ and $v^2/r$, the allowed region is a circle of radius $\mu_s g$, the friction circle. The fastest speed that holds while braking at $a_t$ on a flat curve follows by solving for v:
$v_{\mathrm{max}} = \sqrt[4]{r^2\left((\mu_s g)^2 - a_{t}^2\right)}$
Braking in a curve therefore lowers the cornering limit, and a car that would have held at a steady speed can skid as soon as the driver brakes. The advice in every driving course follows: brake before the curve, not in it. Subtracting the braking from $\mu_s g$ directly, instead of as squares, is the common error; it overstates how much grip braking costs.
On a banked road the budget tilts with the road, and the window of Part 1 gets its exact form.
The exact banking inequality and its limits
Take a car on a curve banked at θ, with the normal force N perpendicular to the road and the friction f along it, counted positive down the bank, toward the center. Its acceleration is horizontal, $v^2/r$ toward the center, so the horizontal and vertical equations are
$N\sin\theta + f\cos\theta = mv^2/r$ $N\cos\theta - f\sin\theta = mg$
Solving them gives $N = m(g\cos\theta + (v^2/r)\sin\theta)$ and $f = m((v^2/r)\cos\theta - g\sin\theta)$. Friction vanishes at the design speed and changes sign there. Static friction holds while $|f| \le \mu_s N$. With $a = v^2/r$, that condition is
$|a\cos\theta - g\sin\theta| \le \mu_s(g\cos\theta + a\sin\theta)$
the test that always applies. Writing $\mu_s = \tan\varphi$ and dividing through gives a window, valid while θ + φ < 90°:
$\tan(\theta - \varphi) \le v^2/(rg) \le \tan(\theta + \varphi)$ ${\tan\varphi = \mu_s}$
with the lower edge zero when θ ≤ φ. The fast side divides by $\cos\theta - \mu_s\sin\theta$: once θ + φ ≥ 90° that is not positive, every speed passes, and the tangent (negative past 90°) is no bound.
The friction angle φ is the half-angle of the friction cone: the road's total push, normal force plus friction, may tilt up to φ from the normal either way. The inequality is then simply a frictionless bank whose angle can be anything from θ − φ to θ + φ.
The limits are worth reading off. If θ ≤ φ, the lower bound is zero or negative: a stopped car can park on the bank. If θ + φ ≥ 90°, there is no upper bound: the harder the car presses into the road, the more friction it gets. As $\mu_s \to 0$, both bounds close on tan θ and only the design speed holds. The mass never appears.
Engineers write the same physics in a form suited to designing roads, with a deliberate margin built in.
How highway engineers bank a curve
Highway designers describe a bank by its superelevation $e = \tan\theta$, the rise across the road per unit width, usually given as a percentage. They describe friction with a side-friction factor f, the friction a car at the design speed is allowed to use, chosen for comfort and far below what the tires could give. Putting $e = \tan\theta$ and $f = \tan\varphi$ into the upper edge of the banking inequality gives
$v^2/gR = (e + f)/(1 - ef) \approx e + f$
Since e and f are both small, their product is tiny, and the simplified form is close. Solved for the radius, it sets the tightest curve allowed for a given design speed:
$R_{\mathrm{min}} = \dfrac{v^2(1 - ef)}{g(e + f)}$
The side-friction factor is not the grip. It limits how hard passengers are pushed sideways, and the gap between f and the tires' $\mu_s$ is a reserve for rain, worn tires and braking in the curve. Typical design values, used here as illustrations, are e up to about 0.08 and f about 0.12 at highway speeds. The common slip is to put the speed in km/h straight into the formula; convert to m/s first.
The conical pendulum of Part 1 hides a simple rule for its own spin, which the next section uncovers.
The cone's spin rate
For the conical pendulum, write the radius of the ball's circle as $L\sin\theta$ for a string of length L, and the speed as ωr. The horizontal equation $T\sin\theta = m\omega^2 L\sin\theta$ then gives $T = m\omega^2 L$, and the vertical one, $T\cos\theta = mg$, turns into
$\cos\theta = g/(L\omega^2)$ $\omega = \sqrt{g/(L\cos\theta)}$
Since cos θ cannot exceed 1, there is no cone at all below $\omega = \sqrt{g/L}$: a ball turned more slowly just hangs straight down and spins on the spot. Faster, the cone opens toward 90° without reaching it. The quantity $L\cos\theta$ is the cone's height h, from the pivot down to the plane of the circle, so $\omega = \sqrt{g/h}$ and the time per turn is $2\pi\sqrt{h/g}$: two cones of the same height turn at the same rate, whatever their strings' lengths. The usual algebra slip puts cos θ in the numerator, as $\sqrt{g\cos\theta/L}$.
A wall can do what a string does, and a spinning drum makes the rider the bob.
The rotor ride
In the rotor, riders stand against the wall of a drum of radius r. The drum spins at ω, and then the floor drops away. In the ground frame the forces are simple. The wall pushes each rider inward with a normal force, and since the wall's push is the only horizontal force, it is the whole net force: $N = m\omega^2 r$. Static friction between the rider's clothes and the wall points up and must hold the weight, at most $\mu_s N$. The rider stays up if $\mu_s m\omega^2 r \ge mg$:
$\omega_{\mathrm{min}} = \sqrt{g/(\mu_s r)}$
The mass cancels, so a heavy rider and a light one slide, or hold, at the same spin. Riders feel pressed outward against the wall, but in the ground frame the wall is pushing them inward, and without it they would leave along the tangent. Spin rates are often quoted in revolutions per minute; one revolution is 2π rad, so multiply rad/s by 60/2π.
Back to the vertical circle, now at any point on it, and with every kind of support.
Anywhere on the vertical circle, and what each support can do
Measure the position on a vertical circle by the angle α from the bottom. Gravity then has two parts: $mg\cos\alpha$ along the radius, pointing outward in the lower half and inward in the upper half, and $mg\sin\alpha$ along the path, which slows the body on the way up and speeds it on the way down. The support (a string, a rod, a track) must supply whatever inward force gravity's radial part does not:
$T = m(v^2/r + g\cos\alpha)$ $a_t = -g\sin\alpha$
At the bottom (α = 0) this is $m(v^2/r + g)$, at the top (α = 180°) $m(v^2/r - g)$, and at the side (α = 90°) gravity is entirely along the path, so the string supplies $mv^2/r$ alone while the speed changes at the full rate g.
What a support can do decides what is possible. A string only pulls toward the center, and a loop's track only pushes toward it, so both need $v^2 \ge gr$ at the top. A rod can push or pull, so it can hold a ball up at the top with no minimum speed. A hill crest only pushes outward, so a car stays on it only while $v^2 \le gr$. The mistake to avoid is treating every support as if it could push and pull.
The banked turn has one more famous case, where the road is the air itself.
A pilot's level turn
An airplane turns by banking its wings. The lift L acts perpendicular to the wings, so in a bank of angle β it tilts toward the center of the turn. In a level turn, at constant altitude, the vertical part of the lift holds up the weight and the horizontal part turns the airplane:
$L\cos\beta = mg$ $L\sin\beta = mv^2/r$
This is the frictionless banked road again, with lift in place of the normal force. Two results follow. The load factor, the lift as a multiple of the weight, depends only on the bank, and the radius of the turn depends on the bank and the speed:
$n = 1/\cos\beta$ $r = v^2/(g\tan\beta)$
At a 75° bank, n = 1/cos 75° = 3.86: everyone aboard feels nearly four times as heavy, and the pilot must pull back to nearly quadruple the lift or the airplane sinks. A faster airplane at the same bank has the same load factor but turns wider. Pilots often fly a standard-rate turn, 3° of heading per second; its bank grows with airspeed. Using sin β where the ratio calls for tan β is the usual slip.
Every result so far treated the moving body as a point. The last section shows where that stops working.
Where the point mass fails: rollover
A real vehicle has a height. Its center of mass sits a height h above the road, and its wheels are a track width t apart. On a flat curve, friction at the tires pushes toward the center, and that push acts at road level, below the center of mass. Taking torques about the center of mass, the friction tends to tip the vehicle toward the outside of the curve, and the normal force shifts to the outer wheels to balance it. The inner wheels' share reaches zero, and they lift, when the friction needed per unit normal force reaches the static stability factor:
${|f|}/N = t/2h$
Sliding needs $|f|/N = \mu_s$, and on a flat road the ratio needed is $v^2/(rg)$, so whichever threshold is lower comes first. If $t/2h > \mu_s$ the tires slide first; if $t/2h < \mu_s$ the vehicle rolls over first. Cars have $t/2h$ of about 1.3 to 1.5 and SUVs about 1.0 to 1.3, so on ordinary roads both slide first; tall, narrow vehicles can fall below a road's $\mu_s$. The condition contains no mass. This rigid model leaves out the body's lean on its springs, which lowers the real threshold, and the tires' slip angles.
Reference
Common mistakes
- Constant speed, no acceleration. A turning velocity is a changing velocity: $a = v^2/r$, toward the center.
- An outward force. In the ground frame nothing pushes outward; a released body leaves along the tangent.
- A "centripetal force" on the diagram. It is a direction. A real force with an agent supplies it.
- Forgetting a square. $v^2/r$, not v/r; and $v_{\mathrm{max}} = \sqrt{\mu_s gr}$, not $\mu_s gr$.
- Kinetic friction in a turn. Tires that roll without sliding grip with static friction, toward the center.
- The ramp's normal force on a bank. At the design speed $N = mg/\cos\theta$, more than the weight; in general $N = m(g\cos\theta + (v^2/r)\sin\theta)$.
- Too slow is safe. On a steep, slippery bank a slow car slides down toward the inside.
- Adding the weight at the top. At the top gravity points to the center: $T + mg = mv^2/r$.
- Adding perpendicular parts as numbers. $a_t$ and $v^2/r$ combine by Pythagoras, in the acceleration and in the friction circle.
- Every support can push and pull. A string only pulls; a loop's track and a hill crest only push.
Equation sheet
| Part 1 | |
|---|---|
| acceleration | $a = v^2/r = \omega^2 r$, $v = 2\pi r/T$ |
| net force | $\Sigma F = mv^2/r$, toward the center |
| string | $T = mv^2/r$ |
| cone | $T\cos\theta = mg$, $T\sin\theta = mv^2/r$ |
| flat curve | $v_{\mathrm{max}} = \sqrt{\mu_s gr}$ |
| bank, no friction | $\tan\theta = v^2/(rg)$, $N = mg/\cos\theta$ |
| window | ${v^2 = rg\tan(\theta \pm \varphi)}$ while θ + φ < 90°; the lower edge at least 0 |
| vertical circle | top: $T + mg = mv^2/r$; slowest: $v = \sqrt{gr}$ |
| Part 2 | |
|---|---|
| vector | $\vec a = -\omega^2\vec r$; ${|a|} = \sqrt{a_{t}^2 + (v^2/r)^2}$ |
| friction circle | $\sqrt{a_{t}^2 + (v^2/r)^2} \le \mu_s g$ |
| bank | ${|a\cos\theta - g\sin\theta| \le \mu_s(g\cos\theta + a\sin\theta)}$, $a = v^2/r$; while θ + φ < 90°: ${\tan(\theta - \varphi) \le v^2/(rg) \le \tan(\theta + \varphi)}$, ${\tan\varphi = \mu_s}$ |
| highway | $v^2/gR = (e + f)/(1 - ef) \approx e + f$ |
| cone | $\cos\theta = g/(L\omega^2)$ |
| rotor | $\omega_{\mathrm{min}} = \sqrt{g/(\mu_s r)}$ |
| side of the circle | $T = m(v^2/r + g\cos\alpha)$ |
| level turn | $n = 1/\cos\beta$, $r = v^2/(g\tan\beta)$ |
| rollover | wheels lift when ${|f|}/N = t/2h$ |
Symbols: v speed · r radius (R for a road) · T period, or tension · ω angular speed (rad/s; 1 rpm = 2π/60 rad/s) · $a_c$, $a_t$ centripetal and tangential acceleration · $\mu_s$ coefficient of static friction · θ bank or cone angle · φ friction angle · N normal force · f friction (+ down the bank) · e superelevation, f side-friction factor · L string length or lift · α angle from the bottom · β bank of a wing · n load factor · t track width · h height of the center of mass · g = 9.81 m/s².
Further reading
OpenStax, University Physics Volume 1, §4.4 (uniform circular motion) and §6.3 (centripetal force), for another account of the same ideas. The friction coefficients come from the library's shared table of approximate values typical of introductory physics tables; the road surfaces, the highway design values and the vehicles' dimensions are illustrative. The Field Manual lists every assumption the model makes.
Circular Motion & Banked Turns v1.6 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org