Inclined Planes & Friction · Guide · glidepathscience.org
GuideInclined Planes & Friction
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. Each section ends with a worked example that is not one of the sim's own problems. Most of them use one pair of surfaces that is not in the sim's table, with $\mu_s$ = 0.65 and $\mu_k$ = 0.45, so you can check them in the Sandbox by choosing Custom.
Part 1 · The essentials
Weight on a slope, and the tilted axes
A crate waiting on a loading ramp, a car parked on a hill and a skier at the top of a run all face the same situation: gravity pulls straight down, but the surface, and everything that can happen along it, is tilted. A ramp is the simplest place to learn how friction works, because the motion is confined to one line while the forces are not. Three forces act on a block resting on a ramp: its weight mg, pulling straight down; the normal force N, the ramp's push perpendicular to its surface; and friction, the ramp's push along its surface.
The trick that makes ramp problems easy is to choose axes that follow the surface: x along the slope, up the slope positive, and y perpendicular to it. The block can only move along x, so the problem becomes one-dimensional. The normal force and friction already lie along these axes; only the weight does not, so we split it in two. The angle between the weight and the perpendicular to the ramp equals the ramp's own angle θ, because the two are measured from lines that are themselves perpendicular. The weight therefore has a part along the slope and a part into the surface:
$W_{\parallel} = mg\sin\theta$ $W_{\perp} = mg\cos\theta$
Here m is the block's mass, g = 9.81 m/s² is the strength of gravity near Earth's surface, and θ is the ramp's angle above the horizontal. $W_{\parallel}$ points down the slope and $W_{\perp}$ presses the block into the surface. Two quick checks keep the trigonometry straight: on a flat ramp, θ = 0, nothing should pull along the surface, and sin 0 = 0; on a vertical wall, θ = 90°, the whole weight pulls along it, and sin 90° = 1.
Across the slope nothing accelerates, so the ramp pushes back exactly as hard as the block presses on it:
$N = mg\cos\theta$
The normal force is perpendicular to the surface, never straight up, and on a slope it is always smaller than the weight, falling toward zero as the ramp nears vertical. Writing N = mg is the most common slip in ramp problems, and it spoils everything that follows, because friction depends on N. A push that presses the block into the surface raises N, and a pull that lifts it lowers N.
Something must balance those 27.6 N along the slope, or the crate would slide. That something is static friction, and it behaves unlike any force we have met so far.
Static friction pushes back only as hard as it must
Two surfaces pressed together resist sliding over each other. As long as they do not slip, the resisting force is static friction, $f_s$. Its most important property is that it has no fixed size. It takes whatever value keeps the surfaces from slipping, up to a ceiling set by how hard they are pressed together:
$f_s \le \mu_s N$
The coefficient of static friction $\mu_s$ is a pure number that belongs to the pair of surfaces in contact, not to either one alone: the sim's table gives 0.50 for wood on wood, 1.0 for rubber on concrete and 0.15 for lubricated metal. Its direction is as adaptable as its size: it opposes the slipping that would happen without it. A block left alone on a ramp would slide down, so friction points up the slope, and it is exactly as large as the pull it balances, $f_s = mg\sin\theta$. A hand that pushes the block up the slope harder than $mg\sin\theta$ would make it slip upward, so friction turns around and points down the slope. A push of exactly $mg\sin\theta$ leaves friction nothing to do, and it is zero.
The usual mistake is to treat $\mu_s N$ as the friction force. It is only the most that static friction can give. A related mistake is to say that friction "opposes motion": a block at rest has no motion to oppose. Friction opposes the slip that the other forces would cause.
What happens as the ramp gets steeper and that 82 % creeps toward 100 %?
The slipping angle
The simplest friction experiment needs no scale and no force gauge, only a protractor: set a block on a board and tilt the board slowly until the block lets go. As the board tilts, two things change at once. The pull down the slope, $mg\sin\theta$, grows, and the ceiling, $\mu_s mg\cos\theta$, shrinks, because a steeper board presses on the block less. Static friction keeps pace with the pull, so for a long while nothing happens at all. Then the growing pull meets the falling ceiling, and at that instant
$mg\sin\theta = \mu_s mg\cos\theta \quad\Rightarrow\quad \tan\theta = \mu_s$
This angle is the slipping angle of the pair, also called its angle of friction or angle of repose (a sand pile's angle of repose is a different, granular effect). Below it the block holds, which is the test for any block left alone on a ramp: it stays put while $\tan\theta \le \mu_s$. Above it the block slides.
The mass and g have both canceled. A block twice as heavy is pulled down the slope twice as hard, but it also presses into the board twice as hard, which doubles the ceiling. On the Moon, where g is 1.62 m/s², both sides shrink by the same factor. Students often expect the heavier block, or the block on Earth, to slip first; in fact four blocks of two masses on two worlds all let go together. The slipping angle measures $\mu_s$ for the pair of surfaces and nothing else. Two other slips are worth avoiding: the test uses $\mu_s$, not $\mu_k$, and it compares $\mu_s$ with the tangent of the angle, never with the angle itself.
The moment the block lets go, a different kind of friction takes over.
Sliding: kinetic friction
Once surfaces slide over each other, friction stops adjusting. Kinetic friction has a fixed size, set by the normal force, and it always points against the sliding:
$f_k = \mu_k N$
The coefficient of kinetic friction $\mu_k$ is usually smaller than $\mu_s$ for the same pair, 0.20 against 0.50 for wood on wood: keeping something sliding is easier than starting it. For a block sliding down a ramp, gravity's pull and kinetic friction point opposite ways along the slope, and Newton's second law gives
$ma = mg\sin\theta - \mu_k mg\cos\theta$ $a = g(\sin\theta - \mu_k\cos\theta)$
where a is the acceleration down the slope. The mass cancels again. In this model $\mu_k$ does not depend on speed, so the acceleration is constant all the way down, and an a(t) graph is a flat line. Three errors are common here: using $\mu_s$ for a block that is already sliding, adding friction to gravity's pull when it should be subtracted, and using mg for the normal force instead of $mg\cos\theta$.
The acceleration is zero when $\tan\theta = \mu_k$. At that angle a nudged block slides down at a steady speed. Because $\mu_k$ is smaller than $\mu_s$, this angle lies below the slipping angle, and between the two the ramp has a memory: a block set down at rest stays put, while one that is already sliding keeps going and even speeds up.
Gravity and friction need not act alone. Often a hand joins in.
Pushing up the slope, and holding with a push
Loading a crate means pushing it up a ramp, and the push needed depends on which way friction points. Pushed up along the slope at a constant speed, the crate has zero acceleration, so the forces along the slope balance. The crate slides up, so kinetic friction points down the slope, alongside gravity's pull, and the push P must beat both:
$P = mg(\sin\theta + \mu_k\cos\theta)$
The speed does not appear: any constant speed takes the same push. Lowering the same crate at a constant speed reverses friction, which now helps the hand, and the push falls to $mg(\sin\theta - \mu_k\cos\theta)$. On a ramp shallower than $\tan^{-1}\mu_k$ that is negative: the crate would stop by itself, and keeping it sliding down takes a push down the slope.
Holding a crate still is different, because static friction can point either way. Suppose the ramp is steeper than the slipping angle, so the crate would slide down on its own. The smallest push that holds it is the one at which friction helps, up the slope at its full $\mu_s N$. The largest is the one at which friction resists, down the slope at its full $\mu_s N$. Every push in between also holds the crate, with static friction taking up whatever is left over:
$mg(\sin\theta - \mu_s\cos\theta) \le P \le mg(\sin\theta + \mu_s\cos\theta)$
At $P = mg\sin\theta$, in the middle, friction is zero. Forgetting that friction flips direction across this band is the common error, along with using $\mu_k$ for a crate that is not moving.
What if no hand pushes at all, and the block is simply given a shove up the slope?
Up, stop and back
A block launched up a rough ramp shows more clearly than anything else that friction depends on the direction of sliding. On the way up, gravity's pull and kinetic friction both point down the slope, so the block slows at
$a = g(\sin\theta + \mu_k\cos\theta)$
With a constant deceleration, $v^2 = v_0^2 - 2ad$, where $v_0$ is the launch speed and d the distance along the slope. The block stops when v = 0, after
$d = v_0^2/(2a)$
At the top its velocity is zero, but its acceleration is not: gravity still pulls it down the slope. What happens next is decided by static friction, exactly as for a block set down at rest. If $\tan\theta \le \mu_s$ the block stays at the top for good. If not, it slides back, and now friction points up the slope, so the acceleration on the way down, $g(\sin\theta - \mu_k\cos\theta)$, is smaller than on the way up. A graph of velocity against time is two straight pieces: a steep one down to zero, then a gentler one below zero. The block passes its launch point slower than it left, and the missing kinetic energy has become thermal energy, $\mu_k N$ times the total distance slid. Expecting the same acceleration both ways, or a return at the launch speed, means leaving out friction's change of direction.
The same reasoning, run backward, lets us design a ramp that never lets a crate go.
Designing a ramp that holds
A ramp built for crates to rest on cannot rely on one value of $\mu_s$: dust, damp, wear and packaging spread the real values over a range, and the ramp must hold the slipperiest crate it will ever see, not the average one. On top of that, a designer keeps a margin, a factor of safety F, by counting on only $\mu_s/F$ of the worst grip. The steepest safe angle and, for a rise h, the shortest ramp are then
$\tan\theta \le \mu_{s,min}/F$ $L = h/\sin\theta$
where $\mu_{s,min}$ is the lowest coefficient in the range and L is the ramp's length along its surface. The common mistakes are designing for the grippiest or the average crate, multiplying by F instead of dividing, and dividing the angle by F instead of the coefficient.
Part 2 returns to the contact between block and ramp with one picture that contains all of these results, and then pushes the model to its limits.
Part 2 · Going further
The friction cone, and self-locking
The normal force and friction are really two components of one contact force. Since $f \le \mu_s N$, the contact force can lean away from the surface's normal by at most the angle whose tangent is $\mu_s$, the friction angle. Every force the surface can exert on a body at rest therefore lies inside a cone of half-angle $\tan^{-1}\mu_s$ around the normal: the friction cone. A body stays at rest only if the contact force it needs lies inside that cone.
For a block alone on a ramp, the contact force must cancel the weight, so it must point straight up. The cone tilts with the ramp, and straight up leaves the cone when the tilt reaches $\tan^{-1}\mu_s$: the slipping angle, now seen as geometry.
The cone also explains a surprising result. Push a block horizontally, toward the slope, to start it up the ramp. Part of every push, $P\sin\theta$, presses the block into the surface and raises the normal force to $mg\cos\theta + P\sin\theta$, which raises the friction to be beaten. The block starts up the slope only when
$P\cos\theta > mg\sin\theta + \mu_s(mg\cos\theta + P\sin\theta)$
so the push needed is
$P = mg(\sin\theta + \mu_s\cos\theta)/(\cos\theta - \mu_s\sin\theta)$
The denominator falls to zero when $\tan\theta = 1/\mu_s$, and beyond that no push, however large, can start the block, since each extra newton adds more to friction's ceiling than to the push along the slope:
$\tan\theta \ge 1/\mu_s$
This is self-locking. In the cone picture, the push now lies within $\tan^{-1}\mu_s$ of the inward normal, so its reverse lies inside the cone, and the surface can cancel any amount of it. Wedges, screws and door stops hold under load for the same reason. The trap is to confuse this angle with the slipping angle: the two are complementary, since $\tan^{-1}(1/\mu_s)$ = 90° − $\tan^{-1}\mu_s$.
If a horizontal push is the worst direction, which direction is the best?
The angle of least effort
Pull a crate up a ramp with a rope at an angle φ above the ramp's surface. Tilting the rope upward does two things: it lifts part of the crate's weight off the surface, lowering N to $mg\cos\theta - P\sin\varphi$ and with it the friction, and it wastes part of the pull, since only $P\cos\varphi$ acts along the slope. Balancing the forces for a steady slide up the slope gives
$P = mg(\sin\theta + \mu_k\cos\theta)/(\cos\varphi + \mu_k\sin\varphi)$
The pull is least where the denominator is largest. Setting its derivative to zero, $-\sin\varphi + \mu_k\cos\varphi = 0$, gives the angle of least effort, and the denominator's largest value is $\sqrt{1 + \mu_k^2}$:
$\tan\varphi^{\text{*}} = \mu_k$ $P = mg(\sin\theta + \mu_k\cos\theta)/\sqrt{1 + \mu_k^2}$
The best angle is measured from the ramp's surface, and it does not depend on the ramp's tilt at all, only on the sliding friction. Measuring it from the horizontal, or using $\mu_s$ for a crate that is already sliding, are the usual errors. The saving is modest when $\mu_k$ is small and grows with it.
We now go back to the moment of the slip, where static friction gives way to kinetic friction.
The lurch at the slip, and stick-slip
On a slowly tilted board, static friction holds the net force at zero right up to the slip. At the slip it drops at once from $\mu_s N$ to $\mu_k N$, and the block lurches away with the acceleration that the difference leaves behind. Using $\tan\theta_s = \mu_s$ at the slipping angle $\theta_s$,
$a = g\cos\theta_s\,(\mu_s - \mu_k)$
If the two coefficients were equal, the block would not lurch at all: it would sit on the edge, and a nudge would send it off at a constant speed. A large lurch is the signature of a large drop from static to kinetic friction.
The same drop drives stick-slip. Drag a block through a spring of stiffness k whose far end moves slowly and steadily. The spring stretches until its pull beats gravity's pull down the slope plus static friction's ceiling, $\mu_s N$. Then the block slips, only kinetic friction resists, and the block surges forward: half an oscillation on the spring, about a center $(\mu_s - \mu_k)N/k$ ahead of where it started. It swings as far beyond that center as it started behind it, stops, and sticks again. In the slow-pull limit each slip covers
$\Delta s \approx 2(\mu_s - \mu_k)\,mg\cos\theta/k$
and lasts half a period, $\pi\sqrt{m/k}$. (A faster pull makes each slip a little longer, since the far end moves during it.) Forgetting the factor of 2, or taking a full period, are the common mistakes. In this model the sawtooth never smooths out, at any pull speed; the last section explains why real stick-slip often does.
Every result so far depends on $\mu_s$ and $\mu_k$, and those have to be measured.
Measuring the coefficients honestly
The slipping angle measures $\mu_s$, but real surfaces never let go at exactly the same angle twice. A careful experimenter repeats the tilt, takes the mean of $\tan\theta$, and reports its standard error: the standard deviation of the trials divided by the square root of their number.
For $\mu_k$, release the block from rest on a ramp steeper than its slipping angle, or start it with a light tap on any ramp steeper than $\tan^{-1}\mu_k$, and time it over a measured distance d. Constant acceleration from rest gives $a = 2d/t^2$. One slide gives one acceleration, but slides at several angles give much more, because dividing $a = g(\sin\theta - \mu_k\cos\theta)$ by $\cos\theta$ turns it into a straight line:
$a/\cos\theta = g\tan\theta - \mu_k g$
Plotted against $\tan\theta$, the points fall on a line whose slope should be g and whose intercept is $-\mu_k g$, so $\mu_k$ = −intercept/slope; the slope checks the experiment. The timing sets the precision. Because the time is squared in $a = 2d/t^2$, its relative error counts twice:
$\sigma_{\mathit{a}}/a = 2\sigma_{\mathit{t}}/t$
A hand stopwatch scatters by about 0.1 s, the variation in a person's reaction time, which ruins short slides; photogates scatter by about half a millisecond. A fit that weights each point by its own uncertainty lets the slow, precise slides count most. Forgetting the factor of 2, reading the intercept as $\mu_k$ without dividing by the slope, and measuring at a single angle are the usual errors.
Not every body slides. A ball or a cylinder can roll, and then friction does something else entirely.
Rolling: friction that does no work
A ball rolling without slipping touches the ramp at a point that is momentarily at rest, so the friction there is static. Gravity acts at the center and has no torque about it, so only friction can change the ball's spin. Write the body's moment of inertia as $I = \beta mr^2$, where r is its radius and β depends on its shape: ⅖ for a solid sphere, ½ for a solid cylinder and 1 for a hoop. Newton's second law for the center, together with the torque equation for the spin and the rolling condition, gives
$a = g\sin\theta/(1 + \beta)$
The friction needed to keep the contact point from slipping is $\beta mg\sin\theta/(1 + \beta)$, and it must not exceed $\mu_s mg\cos\theta$, so the body rolls without slipping only if
$\mu_s \ge \tan\theta\cdot\beta/(1 + \beta)$
Neither result depends on the mass or the radius, only on β. In a race down one ramp, the solid sphere beats the solid cylinder, which beats the hoop, whatever their sizes. A sliding block can finish anywhere among them, because its acceleration, $g(\sin\theta - \mu_k\cos\theta)$, depends on $\mu_k$ instead of β.
Friction on a rolling body defies two common expectations. First, it points up the slope whether the body rolls down or up: rolling up, gravity slows the center faster than anything slows the spin, so the contact point would slip down the slope, and rolling down, the spin must speed up. Either way friction pushes up the slope; "friction opposes motion" would point it down on the way up, and that is wrong. Second, static friction on a body rolling without slipping does no work, because the point where it acts does not move. It makes no heat; it only moves energy between the motion of the center and the spin. A ball rolled up a ramp therefore comes back to its starting point at the speed it left, unlike a sliding block.
Every result in this Guide rests on one simple model of friction. The last section asks where it ends.
Where the friction model ends
The model used throughout is the Amontons–Coulomb model: friction is proportional to the normal force, independent of the apparent area of contact, and, once sliding, independent of speed. The independence of area has a microscopic explanation. Even polished surfaces touch only at the tips of microscopic bumps, squashed until they carry the load. The total real area of contact is set by the load and by the hardness of the softer material, so it grows with the load and not with the size of the face. A narrower face changes nothing (Bowden and Tabor). Rubber is the main exception: it deforms into the surface, and its grip depends on the pressure, the compound and the temperature, so a wider tire may grip more, but not always.
A tall block raises another question: it can tip over before it slides. A uniform block tips when its weight's line of action passes outside its base, which happens when $\tan\theta$ exceeds its width divided by its height. It slides first only if $\mu_s$ is smaller than that ratio. Tipping is a question of torque, which this sim leaves out by drawing squat blocks.
Real friction also changes with time and speed. Surfaces left in contact slowly creep and bond, so $\mu_s$ grows roughly with the logarithm of the time at rest (aging), and kinetic friction usually falls as the sliding speeds up (velocity weakening). Together they drive the stick-slip of squealing brakes, squeaking doors and earthquake faults, and above a critical pulling speed they let it give way to steady sliding, which the model's sawtooth never does. Lubrication changes everything, often lowering μ by a factor of five or more. Above all, μ belongs to a pair of surfaces in a given condition, never to one material: thinking of "the μ of wood" is the last of the common mistakes.
Reference
Common mistakes
- Writing N = mg on a ramp. Across the slope $N = mg\cos\theta$, less than the weight.
- Swapping sine and cosine. Along the slope goes with sin θ: zero on a flat ramp.
- Taking $\mu_s N$ as the friction on a block at rest. It is only the ceiling; static friction matches the pull.
- Saying friction always points down the slope, or against the motion. It opposes the slip that would happen without it.
- Expecting a heavier block, or one on Earth, to slip first. m and g cancel: $\tan\theta = \mu_s$.
- Using $\mu_s$ for a sliding block, or $\mu_k$ for one at rest.
- Expecting the same acceleration up and down. Friction changes direction at the top.
- Designing for the average grip. A ramp must hold the slipperiest crate, with $\mu_s/F$.
- Measuring the best pulling angle from the horizontal. $\tan\varphi = \mu_k$, from the surface.
- Pointing friction down the slope on a ball rolling up. It points up the slope both ways and does no work.
Equation sheet
| Part 1 | |
|---|---|
| along the slope | $mg\sin\theta$ (down); $N = mg\cos\theta$ |
| at rest | $f_s \le \mu_s N$, so it holds while $\tan\theta \le \mu_s$ |
| sliding down | $a = g(\sin\theta - \mu_k\cos\theta)$; steady at $\tan\theta = \mu_k$ |
| sliding up | $a = g(\sin\theta + \mu_k\cos\theta)$ (slowing) |
| pushed up | $P = mg(\sin\theta + \mu_k\cos\theta)$ at constant speed |
| held by a push | $mg(\sin\theta - \mu_s\cos\theta) \le P \le mg(\sin\theta + \mu_s\cos\theta)$ |
| stopping | $d = v_0^2/(2a)$ |
| design | $\tan\theta \le \mu_{s,min}/F$; $L = h/\sin\theta$ |
| Part 2 | |
|---|---|
| least pull | $\tan\varphi^{\text{*}} = \mu_k$, $P = mg(\sin\theta + \mu_k\cos\theta)/\sqrt{1 + \mu_k^2}$ |
| horizontal start | $P = mg(\sin\theta + \mu_s\cos\theta)/(\cos\theta - \mu_s\sin\theta)$; none once $\tan\theta \ge 1/\mu_s$ |
| lurch | $a = g\cos\theta_s\,(\mu_s - \mu_k)$ |
| stick-slip | $\Delta s \approx 2(\mu_s - \mu_k)\,mg\cos\theta/k$, in $\pi\sqrt{m/k}$ |
| measuring | $a/\cos\theta = g\tan\theta - \mu_k g$; $\sigma_{\mathit{a}}/a = 2\sigma_{\mathit{t}}/t$ |
| rolling | $a = g\sin\theta/(1 + \beta)$, needs $\mu_s \ge \tan\theta\cdot\beta/(1 + \beta)$ |
Symbols: m mass (kg) · g = 9.81 m/s² · θ ramp angle above the horizontal ($\theta_s$ at the slip) · N normal force (N) · $f_s$, $f_k$ static and kinetic friction (N) · $\mu_s$, $\mu_k$ coefficients of static and kinetic friction · a acceleration along the slope (m/s²) · $v_0$ launch speed (m/s) · d distance along the slope (m) · P push or pull (N) · φ rope angle above the surface · F factor of safety · h rise and L length of a ramp (m) · k spring stiffness (N/m) · Δs slip distance (m) · β = I/mr² (sphere ⅖, cylinder ½, hoop 1) · σ an uncertainty.
Further reading
Samuel J. Ling, Jeff Sanny and William Moebs, University Physics Volume 1 (OpenStax, 2016), free online under CC BY-NC-SA 4.0: Section 6.2, Friction, and Section 11.1, Rolling Motion. F. P. Bowden and D. Tabor on the real area of contact, and J. H. Dieterich and A. Ruina on aging and velocity weakening, are cited in full in the Field Manual (H), which also says how the model works and what it leaves out.
Inclined Planes & Friction v1.4 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org