ARCHETYPE: Conveyor Belts

Drop a parcel onto an airport baggage belt and watch closely. For a fraction of a second the box skids — the belt hisses beneath it — and then it is simply riding, moving with the belt as if it had always belonged there. Friction did that. The force you have met mostly as a brake here plays the opposite role: it grabs the slower parcel and accelerates it forward, doing honest positive work, until the slipping stops and the ride begins. That two-phase story — a friction phase while the surfaces slip, then a ride phase once speeds match — is the entire conveyor-belt archetype, and JEE has been dressing it up for decades: parcels dropped from rest, blocks sliding on faster than the belt, blocks launched against the belt's motion, belts on inclines. Every variant is solved by one question asked at one instant: which way does the block slide relative to the belt? Kinetic friction opposes exactly that relative slide — which is why it can point forward, why it does not switch off when the block momentarily stops, and why every "friction opposes motion" reflex fails here on schedule. The archetype also carries mechanics' most instructive energy audit. One friction force, three different distances — the block's, the belt's, and the slip between them — give three different energy entries, and for the from-rest drop the books close with an unforgettable punchline: the motor pays exactly twice the kinetic energy the parcel receives, and precisely half the payment burns at the interface, independent of mass, coefficient, and speed. This article drills both layers: the Newton-and-kinematics machinery of the two phases, the launched-backward variant with its famous non-flip of friction, and the three-ledger energy audit that separates formula-pluggers from auditors.

Key takeaways

A block on a moving belt has two phases: slip (kinetic friction $f = \mu_k mg$ acts while block and belt speeds differ) and ride (speeds matched; kinetic friction switches off, static friction supplies only what the ride demands — zero on a straight horizontal belt).
Friction's direction is set by relative velocity: it opposes the block's motion relative to the belt. A block slower than the belt is dragged forward; a block faster is braked; a block moving backward keeps one friction direction even as its velocity passes through zero.
The slip phase is constant-acceleration kinematics at $a = \mu_k g$ (mass cancels), lasting $t^* = |v_0 - v_b|/(\mu_k g)$ .
Energy audit: work on the block uses the block's displacement; the motor pays through the belt's displacement; heat is $Q = f \times s_{\text{slip}}$ — three distances, three numbers.
From-rest drop: the belt travels exactly twice the block's distance, so the motor pays $mv_b^2$ — half becomes kinetic energy, half burns. A 50/50 split fixed by physics.
Key idea: one question runs the whole archetype — which way does the block slide relative to the belt? Friction opposes that, at $\mu_k mg$ , until the slip dies.

The core insight

Conveyor-belt problems are two-phase motion: kinetic friction $\mu_k mg$ , directed against the block's velocity relative to the belt, accelerates or brakes the block at $\mu_k g$ until speeds match; thereafter static friction takes over with zero demand on a straight horizontal belt — and the heat generated equals the friction force times the relative slip distance, never times either body's own displacement.

Where this lives today

Every second, somewhere, millions of parcels land on moving belts — a single large fulfillment centre runs tens of kilometres of conveyor. The physics above is the design sheet: motors are sized from the double-payment rule (each from-rest loading event costs $mv_b^2$ ), and belt compounds are chosen to survive the concentrated $f\,s_{\text{slip}}$ heating at loading zones — the scuffed, polished patches you can see at any airport carousel mouth. High-speed sorters sidestep the tax entirely with powered induction ramps that match a parcel's velocity to the belt before contact: the engineered $v_0 = v_b$ limit, where slip, heat, and scuffing all vanish. Treadmills run the archetype on your body twice a stride (your foot lands slower than the belt and is dragged up to speed), and regenerative braking runs Example 2's footnote as the main event — the "energy into the belt drive" entry, harvested on purpose in every electric vehicle. Wherever a surface must accept cargo moving at the wrong speed, someone has done this article's bookkeeping first.

The setup

Geometry first, because in an archetype the setup is the physics. A belt runs horizontally at constant speed $v_b$ (a motor holds it there, whatever lands on it). A block of mass $m$ arrives with some initial velocity $v_0$ — zero for the classic drop, faster than the belt, or even opposite to it. Coefficient $\mu_k$ (and $\mu_s$ ) between block and belt. Stand in the ground frame and ask the archetype's one question at touchdown: which way does the block move relative to the belt? For the from-rest drop, the belt streams forward beneath the block, so relative to the belt the block slides backward — and kinetic friction on the block points forward, along the belt's motion. Forward force on a body that then moves forward: friction does positive work, accelerating the block at $a = \mu_k g$ until the speeds match. At that instant the relative slide dies, the kinetic law retires, and static friction takes over — supplying, on a straight horizontal belt at constant speed, exactly nothing, because nothing is demanded. The block rides.

Figure

The catch-up phase — one force, three different distances

Friction is a two-way street, and the belt's side of it matters. By Newton's third law the block drags backward on the belt with the same $f$ ; the motor stiffens and feeds in extra power to hold $v_b$ . So the motor is the paymaster — and the geometric fact that runs the energy audit is drawn above: while the block catches up (averaging half the belt's speed from rest), the belt travels exactly twice the block's distance. The difference — belt distance minus block distance — is the slip: how far the two surfaces scraped past each other, the length of the hiss, the distance that becomes heat.

The math: the two phases and the three ledgers

Phase 1 — slip. While speeds differ, kinetic friction has fixed magnitude and known direction:

f = \mu_k m g, \qquad a = \mu_k g \qquad (1)

where: the direction opposes the block's velocity relative to the belt, and the mass has cancelled from $a$ — every parcel, heavy or light, catches up on the same schedule (the force scales with $m$ , the inertia divides it out — free fall's cancellation, lying down). The phase is plain suvat: from-rest drop matches speeds at

t^* = \frac{v_b}{\mu_k g}, \qquad s_{\text{blk}} = \frac{v_b^2}{2\mu_k g}, \qquad s_{\text{belt}} = 2\,s_{\text{blk}}, \qquad s_{\text{slip}} = s_{\text{belt}} - s_{\text{blk}} = \frac{v_b^2}{2\mu_k g} \qquad (2)

where: the factor of two is exact for any from-rest drop — full speed versus average half speed. For a general arrival $v_0$ (same direction), replace $v_b$ by the closing speed $|v_0 - v_b|$ in $t^*$ and compute displacements honestly from suvat; the slip is always (faster body's distance) − (slower body's distance).

Phase 2 — ride. At matched speeds the relative slide is zero; kinetic friction is off duty by definition. Static friction now answers demand — and on a straight horizontal belt at constant velocity the block needs zero net force, so the friction is zero (not $\mu_s mg$ ). The belt could carry the parcel to the horizon force-free. Static friction re-awakens only if the ride itself demands force: the belt accelerates, curves, or climbs.

Derivation AThe launched-against-the-belt block — and why friction refuses to flip at v = 0

We solve the JEE-Main favourite: a block launched at $v_0$ against a belt moving at $v_b$ the other way, tracking the friction direction with the relative-velocity rule through the block's turnaround. The payoff is the archetype's sharpest trap: the block's velocity passes through zero, and the friction does NOT flip there. Watch the one question — relative velocity — asked at every stage.

Step 1. Set positive $x$ along the belt's motion. Belt: $+v_b$ . Block at launch: $-v_0$ . Relative velocity of block with respect to belt: $-v_0 - v_b$ — negative, i.e. the block slides backward relative to the belt.

Step 2. Kinetic friction opposes the relative slide: it points forward ( $+x$ ), magnitude $\mu_k mg$ , acceleration $+\mu_k g$ . The block decelerates, stops momentarily, reverses, and speeds up in the belt's direction — all under this ONE unchanged force. Why: at the instant the block's velocity is zero, its velocity relative to the belt is $0 - v_b = -v_b \ne 0$ — the surfaces are still scraping, the kinetic law still applies, and the direction test still says forward. Friction cares about the interface, not the odometer. The reflex "friction flips when the block turns around" is the single most-harvested error in belt problems.

Step 3. The slip phase ends only when the block reaches belt velocity $+v_b$ . Total change of velocity: from $-v_0$ to $+v_b$ , all at $+\mu_k g$ :

t^* = \frac{v_0 + v_b}{\mu_k g}

Step 4. Displacement bookkeeping splits naturally at the turnaround: backward drift $d_1 = \dfrac{v_0^2}{2\mu_k g}$ (launch to stop), forward leg $d_2 = \dfrac{v_b^2}{2\mu_k g}$ (stop to match). Net block displacement $= d_2 - d_1$ — negative (net backward) whenever $v_0 > v_b$ . Verified: limits — $v_b \to 0$ reduces to a block skidding to rest on the ground: $t^* = v_0/\mu_k g$ , distance $v_0^2/2\mu_k g$ ✓; $v_0 \to 0$ recovers the from-rest drop. Numbers: $v_0 = 4$ , $v_b = 2$ m/s, $\mu_k = 0.4$ , $g = 10$ : $a = 4\ \mathrm{m/s^2}$ , $t^* = 1.5$ s, $d_1 = 2.0$ m back, $d_2 = 0.5$ m forward — net 1.5 m backward before the ride begins.

Go deeper · Derivation B · 1 minThe energy audit — three ledgers, one force, and the 50/50 punchline

Derivation BThe energy audit — three ledgers, one force, and the 50/50 punchline

We close the books on the from-rest drop: work done on the block, work the motor pays, heat at the interface. Strategy: multiply the one force $f$ by the three distances of equation (2) — and then check the whole thing again from the belt's frame, where the story changes but the scorch mark doesn't. Watch whose displacement enters each line: that is the entire subtlety.

Step 1. Work on the block — force forward, block's displacement forward:

W_{\text{blk}} = f\,s_{\text{blk}} = \mu_k mg \cdot \frac{v_b^2}{2\mu_k g} = \tfrac{1}{2}mv_b^2

Exactly the block's kinetic-energy gain — the work–energy theorem audited and passing.

Step 2. The motor's bill — the block's third-law drag $f$ acts backward on the belt through the belt's displacement $2s_{\text{blk}}$ ; holding $v_b$ costs

W_{\text{motor}} = f \cdot 2s_{\text{blk}} = m v_b^2

Step 3. Heat — friction between scraping surfaces burns energy at (force) × (relative sliding distance):

Q = f\,s_{\text{slip}} = \mu_k mg\cdot\frac{v_b^2}{2\mu_k g} = \tfrac{1}{2}mv_b^2

Why: only relative scraping makes heat — a force between surfaces moving together dissipates nothing. Notice $\mu_k$ cancels: a grippier belt doubles the force but halves the slip, and the interface burns the same half — the heat is fixed by the mismatch of speeds, not by the roughness.

Step 4. Close the books: $W_{\text{motor}} = \Delta KE + Q$ , i.e. $mv_b^2 = \tfrac{1}{2}mv_b^2 + \tfrac{1}{2}mv_b^2$ — the motor pays double, half rides away as kinetic energy, half burns. And the frame check: ride the belt, and the block arrives sliding backward at $v_b$ , friction (still forward) brakes it to rest — negative work $-\tfrac{1}{2}mv_b^2$ on the block, zero motor work, same $Q$ . Work and KE are frame-dependent bookkeeping; the slip distance — and the heat — is the frame-proof physical fact. Verified: both frames give $Q = \tfrac{1}{2}mv_b^2$ and both balance separately. The 50/50 from-rest split is standard (Kleppner Ch. 5; Sherwood, Am. J. Phys. 51, 597) and re-derived numerically in Example 4.

Figure

The velocity–time graph is the whole problem — areas are the three distances

Which variant, which move

variant	friction direction (on block)	key result
dropped from rest	forward	$a = \mu_k g$ ; ride after $t^* = v_b/\mu_k g$ ; 50/50 energy split
arrives faster than belt (same direction)	backward	brakes to $v_b$ ; excess KE → heat + belt drive
launched against the belt	forward, UNCHANGED through $v = 0$	one acceleration start to finish; net drift usually backward
belt speeds match at arrival	none (static, zero demand)	no slip, no heat — the gentle-landing limit
inclined belt carrying the block up	static, up-slope, demand $mg\sin\theta$	rides iff $\tan\theta \le \mu_s$ — the repose test on a machine

Limiting cases

case	result	meaning
$v_0 = v_b$	no slip, $Q = 0$	matched speeds: nothing scrapes, nothing burns
$\mu_k \to \infty$	$t^* \to 0$ , $Q$ unchanged at $\tfrac12 mv_b^2$	grip shortens the slip, not the bill
$\mu_k \to 0$	block never catches up	ice on steel: slides forever, belt useless
$v_b \to 0$	ordinary skid-to-rest on ground	the belt family contains the plain floor as a member
$m \to 2m$	same $a$ , same $t^*$ ; energies double	schedule is mass-free, the bill is not
belt accelerating during the ride	static friction re-awakens, $f = ma$ up to $\mu_s mg$	the stacked-blocks problem, with rubber instead of a slab

Common trap"Heat generated = friction force × the block's displacement"

Wrong in general, and the belt is where the error gets caught. Friction force times the block's displacement is the work done on the block — its kinetic-energy change. The heat is the force times the slip distance: how far the surfaces scraped past each other, which differs from either body's own path whenever the floor itself moves. In Example 2 below, friction removes $8.0$ J from the block, but only $4.0$ J burns — the other $4.0$ J flows through friction's third-law twin into the belt drive. The two ledgers happen to agree for the block-dropped-from-rest case (slip = block's displacement there, by the factor-of-two geometry), which is exactly how the misconception breeds: it is trained on the one case where it cannot be caught. The audit discipline: name the force, name whose displacement, and reserve $f \times s_{\text{slip}}$ for the heat — in every frame, for every landing.

Worked examples

1 · Board level — parcel on the loading belt (full method)

A $2.0$ kg parcel is placed at rest on a belt moving at $2.0$ m/s; $\mu_k = 0.20$ , $g = 10\ \mathrm{m/s^2}$ . Find the friction force and acceleration during the slip, the time and block-displacement to reach belt speed, and the friction force after that.

Step 1. Relative-velocity question: block at rest, belt forward ⇒ block slides backward relative to belt ⇒ kinetic friction forward: $f = \mu_k mg = 0.20 \times 2.0 \times 10 = \boxed{4.0\ \mathrm{N}}$ ; $a = \mu_k g = \boxed{2.0\ \mathrm{m/s^2}}$ .

Step 2. Catch-up time: $t^* = v_b/a = 2.0/2.0 = \boxed{1.0\ \mathrm{s}}$ .

Step 3. Parcel's displacement: $s = \tfrac{1}{2}at^{*2} = \tfrac{1}{2}(2.0)(1.0)^2 = \boxed{1.0\ \mathrm{m}}$ (belt meanwhile moved 2.0 m — the factor of two on display).

Step 4. After matching: relative slide zero ⇒ kinetic law retired; the straight, constant-speed ride demands nothing ⇒ friction $= \boxed{0}$ (not $\mu_s mg$ — the ceiling isn't a reading).

2 · Board level — the block that outruns the belt

A $1.0$ kg block slides onto a belt at $5.0$ m/s while the belt runs the same direction at $1.0$ m/s; $\mu_k = 0.25$ , $g = 10\ \mathrm{m/s^2}$ . Find the block's motion until the ride begins, the heat generated, and where the block's lost energy went.

Step 1. Block faster than belt ⇒ slides forward relative to it ⇒ friction backward: deceleration $\mu_k g = 2.5\ \mathrm{m/s^2}$ ; matching after $t^* = (5.0 - 1.0)/2.5 = 1.6$ s.

Step 2. Displacements: block $s_{\text{blk}} = 5.0(1.6) - \tfrac{1}{2}(2.5)(1.6)^2 = 8.0 - 3.2 = 4.8$ m; belt $s_{\text{belt}} = 1.6$ m; slip $= 3.2$ m.

Step 3. Heat: $Q = f\,s_{\text{slip}} = (0.25 \times 1.0 \times 10)(3.2) = \boxed{8.0\ \mathrm{J}}$ ... audit before boxing: block's KE loss $= \tfrac{1}{2}(1.0)(5.0^2 - 1.0^2) = 12$ J; work extracted from block $= f\,s_{\text{blk}} = 2.5 \times 4.8 = 12$ J ✓. Heat $= 2.5 \times 3.2 = \boxed{8.0\ \mathrm{J}}$ .

Step 4. The missing $4.0$ J: the block's forward friction-twin drives the belt through 1.6 m: $2.5 \times 1.6 = 4.0$ J delivered into the belt drive — friction as a crude regenerative brake. Ledger: $12 = 8 + 4$ ✓.

3 · Competitive level — launched against the belt (JEE-Main pattern, ~3.5 minutes)

A block is projected at $4.0$ m/s opposite to the motion of a belt running at $2.0$ m/s; $\mu_k = 0.40$ , $g = 10\ \mathrm{m/s^2}$ . Find (a) the time until the block stops slipping, (b) the block's displacement then, (c) the friction direction at the instant the block's velocity is zero.

Step 1. Positive $x$ along the belt. Block: $-4.0$ m/s; belt: $+2.0$ . Relative velocity $-6.0$ m/s ⇒ friction forward ( $+x$ ), $a = +4.0\ \mathrm{m/s^2}$ , and it stays forward until the block reaches belt velocity.

Step 2. (a) From $-4.0$ to $+2.0$ at $4.0\ \mathrm{m/s^2}$ : $t^* = 6.0/4.0 = \boxed{1.5\ \mathrm{s}}$ .

Step 3. (b) Displacement $= v_0t + \tfrac{1}{2}at^2 = (-4.0)(1.5) + \tfrac{1}{2}(4.0)(2.25) = -6.0 + 4.5 = \boxed{1.5\ \mathrm{m}}$ backward (drifts 2.0 m back to the turnaround, returns 0.5 m).

Step 4. (c) At $v_{\text{block}} = 0$ : relative velocity $= 0 - 2.0 = -2.0 \ne 0$ — still scraping, still kinetic, still forward. No flip. Timing note: one axis, one signed acceleration, one suvat pass — 2.5 minutes if you resist splitting the motion at $v = 0$ (nothing changes there!). Part (c) is the mark-harvester: "friction reverses when the block reverses" is the planted wrong answer. The slip ends at $v = +2.0$ (belt speed), never at $v = 0$ . Budget 3.5 min.

4 · Competitive level — JEE stretch: the motor's bill (~4 minutes)

Sand grains of total mass $3.0$ kg are dropped gently (at rest) onto a belt held at $2.0$ m/s by its motor; $g = 10\ \mathrm{m/s^2}$ . By the time all grains ride the belt, find (a) the kinetic energy given to the sand, (b) the heat generated, (c) the total work the motor supplied for the loading, and (d) show the split survives any $\mu_k$ .

Step 1. (a) Every grain ends at belt speed: $\Delta KE = \tfrac{1}{2}mv_b^2 = \tfrac{1}{2}(3.0)(4.0) = \boxed{6.0\ \mathrm{J}}$ .

Step 2. (b) From-rest drop ⇒ slip distance = block distance ⇒ $Q = f\,s_{\text{slip}} = \tfrac{1}{2}mv_b^2 = \boxed{6.0\ \mathrm{J}}$ .

Step 3. (c) Motor pays through the belt's (doubled) displacement: $W = mv_b^2 = \boxed{12\ \mathrm{J}}$ — and the audit closes: $12 = 6 + 6$ .

Step 4. (d) $\mu_k$ cancelled in Steps 1–3: a grippier belt grabs harder over a shorter slip, a slicker one gentler over a longer slip — force × slip is fixed at $\tfrac{1}{2}mv_b^2$ by the speed mismatch alone. The 50/50 split is physics, not engineering. Timing note: recognize "dropped gently from rest onto a moving belt" and the three answers are $\tfrac{1}{2}mv_b^2$ , $\tfrac{1}{2}mv_b^2$ , $mv_b^2$ on sight — 90 seconds. The (d) concept ("independent of μ") is a standing JEE-Advanced true/false. Continuous-stream versions (sand falling at rate $dm/dt$ ) give the same split as power: belt force $= v_b\,dm/dt$ , motor power $= v_b^2\,dm/dt$ , half burning — the variable-mass module will formalize it. Budget 4 min.

Exam traps

Friction aimed against the block's motion instead of the relative slide. The founding error — the from-rest drop has friction forward. Ask the relative-velocity question first, always.

Flipping friction at $v = 0$ . In the launched-against-the-belt variant, the block's turnaround changes nothing: relative velocity is still nonzero (Example 3c). The slip ends at belt speed, not at rest.

Ride-phase friction quoted as $\mu_s mg$ . After matching, friction equals the demand — zero on a straight horizontal constant-speed belt. The ceiling is a limit, not a reading (the stacked-blocks lesson, verbatim).

Heat computed with the wrong distance. $Q = f \times s_{\text{slip}}$ , never $f \times s_{\text{blk}}$ (that's the block's work) nor $f \times s_{\text{belt}}$ (that's the motor's) — see the pitfall box.

"Doubling $\mu$ doubles the heat." It doesn't touch it: force doubles, slip halves. $Q$ from rest is $\tfrac{1}{2}mv_b^2$ , μ-free — a favourite conceptual MCQ.

Forgetting the motor's double payment. Sizing questions ("work done by the motor per parcel") want $mv_b^2$ , not $\tfrac{1}{2}mv_b^2$ . The belt moves twice as far as the parcel while dragging it up to speed.

Inclined-belt shortcut. A belt carrying a block up a slope at constant speed needs static friction $mg\sin\theta$ (demand), available only if $\tan\theta \le \mu_s$ — the angle-of-repose test wearing a drive motor. No slip means no $\mu_k$ anywhere in it.

Which $g$ . JEE papers declare $g = 10\ \mathrm{m/s^2}$ nearly always; every time and distance in the archetype scales as $1/g$ through $a = \mu_k g$ .

Origins · 1 minGaspard-Gustave de Coriolis and how this was discovered

Gaspard-Gustave de Coriolis

1792–1843 · École Polytechnique, Paris

Coriolis is famous for the deflection force that bears his name, but his deeper legacy sits inside this very archetype. In his 1829 treatise Du calcul de l'effet des machines he pinned down the modern definition of mechanical work — force times the displacement of its point of application — and championed $\tfrac{1}{2}mv^2$ (rather than $mv^2$ ) as its natural companion, so work and kinetic energy would balance cleanly. That precision is exactly what the conveyor belt tests: one friction force does $+\tfrac{1}{2}mv_b^2$ of work on the block, costs the motor $mv_b^2$ , and burns $\tfrac{1}{2}mv_b^2$ at the interface — and only Coriolis's insistence on whose displacement keeps the three from blurring into one. He wrote for engineers sizing waterwheels and steam engines; his ledger discipline now sizes every conveyor motor on Earth.

History

1781

— Coulomb's prize memoir fixes the kinetic-friction law

f = \mu_k N

used in every belt problem.

1829

— Coriolis defines work as force times the displacement of its point of application — the key that makes belt bookkeeping possible.

1843

— Joule measures the mechanical equivalent of heat: the joules friction removes reappear, joule for joule, thermally.

1892

— Thomas Robins develops the troughed conveyor belt for Edison's ore mills; slip heating at loading points becomes an engineering line item.

1966

— Bowden and Tabor's tribology explains microscopically where

f\,s_{\text{slip}}

goes: shearing of micro-welded asperity junctions.

Going further

Texts. Kleppner & Kolenkow Ch. 5 (work–energy with friction) · HC Verma Vol. 1, Ch. 6 · Sherwood, "Pseudowork and real work," Am. J. Phys. 51, 597 (1983) — the classic on exactly this subtlety.

Provenance. All results are standard. A verifier should recompute: the factor-of-two belt/block geometry and the 50/50 split $mv_b^2 = \tfrac{1}{2}mv_b^2 + \tfrac{1}{2}mv_b^2$ ; Example 2's ledger $12 = 8 + 4$ J; Example 3's $t^* = 1.5$ s, net $1.5$ m backward, and the no-flip at $v = 0$ ; Example 4's $6/6/12$ J with the $\mu_k$ cancellation.

Prerequisites. Kinetic & Rolling Friction · Relative Velocity & Reference Frames · Newton's Second Law · Stacked Blocks & Friction Between. Leads to. Which Conservation Law? (technique) · Rocket Propulsion & Variable Mass (the sand-stream limit) · Energy Accounting on Rough Inclines. Deep dive (UG). Friction Forces (UG treatment).

How can friction do positive work? I learned it always removes energy.

Friction opposes relative slipping between surfaces, and if the surface underneath moves forward faster than the body, "opposing the slip" means pushing the body forward — force and displacement aligned, work positive. What friction always does is burn energy at the interface ( $f \times s_{\text{slip}} > 0$ whenever there's slip); what it does to any one body depends on geometry. The belt is the cleanest case: $+\tfrac{1}{2}mv_b^2$ to the block, $-mv_b^2$ from the motor, $+\tfrac{1}{2}mv_b^2$ to heat. Walking and driving run on the same loophole — the ground's friction is the only forward force your body ever gets.

Why doesn't friction switch off (or flip) when the launched-backward block momentarily stops?

Because the kinetic-friction law never looks at the block's velocity — it looks at the relative velocity of the touching surfaces. At the block's turnaround the belt still streams past underneath at full speed; the interface is scraping as hard as ever, so the kinetic force keeps its magnitude and direction. The only special instant in a belt problem is when the block reaches belt velocity: there the scraping dies, and the law changes personality (kinetic → static). Train the reflex: mark the belt's velocity on your diagram and treat "block reaches $v_b$ " as the event, "block reaches 0" as a non-event.

After the speeds match, is the friction static or kinetic, and how big is it?

Static — the surfaces move together — and its size is the demand, not the ceiling. On a straight horizontal belt at constant speed the block needs zero net force to keep riding: friction is exactly zero. It re-awakens the moment the ride demands force: the belt accelerates ( $f = ma$ , up to $\mu_s mg$ — the stacked-blocks threshold), goes around a curve (centripetal demand), or climbs a slope ( $mg\sin\theta$ ). Quoting $\mu_s mg$ for a demand-free ride is the same lost mark as quoting $\mu N$ for a block resting on level ground.

Why is the heat independent of μ? More friction should burn more, surely.

More friction per metre, but fewer metres. The from-rest slip distance is $v_b^2/(2\mu_k g)$ — double $\mu_k$ and the block matches speeds in half the slip. The product $f \times s_{\text{slip}} = \mu_k mg \times v_b^2/(2\mu_k g) = \tfrac{1}{2}mv_b^2$ : the $\mu_k$ cancels. Physically, the heat is fixed by what must be reconciled — the speed mismatch — not by how abruptly the reconciliation happens. The same cancellation puts the total heat of a skidding car at $\tfrac{1}{2}mv^2$ regardless of tyre compound; rubber chemistry chooses where it burns, not how much.

The belt frame gives a totally different work ledger. Which frame is "right"?

Both — for their own questions. Work and kinetic energy are frame-dependent quantities: in the ground frame friction feeds the block $+\tfrac{1}{2}mv_b^2$ ; in the belt frame the block arrives with $\tfrac{1}{2}mv_b^2$ and friction brakes it to rest, work $-\tfrac{1}{2}mv_b^2$ . Each frame's energy books balance internally. The frame-independent entry is the heat, $Q = f\,s_{\text{slip}}$ , because relative slip looks the same to every inertial observer — the scorch mark on the belt is a physical fact. Exam guidance: compute in the ground frame (the motor's role is visible there), and use the belt frame as a 30-second consistency check.

What changes on an inclined conveyor — the ones that carry gravel uphill?

The ride phase stops being free. Carrying a block up at constant speed demands static friction $mg\sin\theta$ up-slope, available only while $\tan\theta \le \mu_s$ — the angle-of-repose test with a motor attached, and the reason real bulk-material belts run at modest angles (or use cleats/buckets beyond them). The landing phase also re-prices: normal force becomes $mg\cos\theta$ , and gravity's along-slope component joins the friction in the acceleration bookkeeping — run the same relative-velocity question on the tilted axis and the incline-family machinery takes over. The physics kit is unchanged: one question for direction, $\mu N$ for magnitude, suvat until the slip dies.