Friction is a branch decision

In the elementary Coulomb model, static friction adjusts within a bounded range to prevent relative tangential slipping at the contact. Its direction opposes the slipping that would occur without it, not necessarily the body's laboratory velocity. A box at rest on an accelerating conveyor may need forward static friction to share the belt's acceleration. A rolling driven wheel can also receive forward static friction from the road. Determine the force needed for the proposed no-slip motion from Newton's law, then compare its magnitude with the maximum permitted by the contact model.

The normal reaction is a separate unknown. On a horizontal floor with no other vertical force and no vertical acceleration, it equals weight. An upward angled pull reduces the normal reaction and therefore reduces the static-friction capacity. On an incline, the perpendicular force balance supplies the normal reaction. Do this calculation before applying the inequality. If the required static force is smaller than the maximum, use the required value itself. Replacing it by the maximum in every sticking situation invents an acceleration that can violate the very no-slip assumption being tested.

Once the surfaces slide relative to one another, a common approximation gives kinetic friction a constant magnitude proportional to normal load. Its direction is opposite the relative tangential sliding velocity at that contact. For a block sliding right on a stationary table, friction points left. If a belt moves right faster than a right-moving block, the block slips left relative to the belt and friction on the block points right. Naming both contacting surfaces and their relative motion removes the ambiguity in the phrase friction opposes motion.

When the relative slip speed reaches zero, the sliding rule alone no longer decides what happens next. Retest the static condition. If the sticking force required at that instant is allowed, the surfaces can stick; otherwise slipping resumes in the direction consistent with the net force and constraint. Numerical solvers need event handling or a carefully chosen regularized contact model here. Simply continuing a constant kinetic force through zero speed can create artificial reversals and chatter. These are failures of branch handling, not evidence that dry friction really supplies an arbitrary oscillating force at rest.

Constant friction coefficients summarize a restricted range of loads, surfaces, speeds, and environmental conditions. They do not follow directly from Newton's laws. Real contacts deform, form microscopic junctions, heat up, wear, and can depend on dwell time or humidity. Apparent contact area can be unimportant in one dry-contact regime and important in another. State which measurements calibrated the coefficient and avoid extending the model automatically to lubricated bearings, rubber tires over every speed, or nanoscale contacts. A simple model can be useful without being a universal law of matter.

Energy transferred out of mechanical motion by dry friction does not disappear. For a sliding block on a stationary rough plane with constant friction magnitude, friction's work is minus that magnitude times the sliding distance; total kinetic-energy change also includes work by gravity and any other forces. A downhill block can therefore speed up while friction dissipates energy. On a moving belt, the motor can also supply energy, and friction's laboratory work on an individual body can be positive. Friction opposes relative slip and produces nonnegative contact dissipation in this ideal model; individual-body work depends on its displacement in the chosen frame. Boundary selection remains as important for energy accounting as for force diagrams.