This page is the toolbox for the Drag topic. Before we can talk about form drag or skin friction, every letter and squiggle in those formulas must mean something you can picture. So we build them one at a time, from absolute zero. Nothing below assumes you have seen the notation before.
Picture a solid object — a ball, a plate, a car — sitting in moving air or water. Or the object moving and the fluid still; it's the same thing, only the point of view changes. The fluid slides past the surface and, everywhere it touches, it presses and rubs. Our whole job is to name each piece of that interaction.
Figure 1 — One tiny surface patch (orange). The blue arrow is the straight-in push (pressure); the green arrow is the sideways rub (shear). Every symbol on this page names an arrow, the patch, or the fluid doing the pushing.
Picture: the length of the long fluid-flow arrows in Figure 1. Longer arrow = bigger v.
Why the topic needs it: the whole reason drag exists is that fluid has to be shoved out of the way, and faster shoving costs more. Drag will turn out to depend on v2 (speed squared), so v is the master dial.
Picture: a small arrow taped onto the free stream far from the body, pointing downstream; the whole x-axis lies along it.
Why we need it: drag is specifically the part of the fluid's push that points downstream along x^. To extract "the part along x^" we will compare other arrows to this one (see §6).
Picture: imagine the fluid packed against the front of the ball, squeezing inward on every tile equally hard from all sides. The denser the squeeze, the bigger p.
Why we need it: the front of a body sits in high pressure (fluid piling up), the back often in low pressure (fluid can't fill the gap). That front-minus-back imbalance is one of the two drags — form drag. Without dFp we couldn't write it down.
This is the trickier of the two forces, so we build it in four steps with a picture.
Figure 2 — The boundary layer. Speed u is zero at the wall (no-slip) and grows with height y. The red dashed line is the slope du/dyat the wall — this steepness alone sets the rub.
Why a derivative and not just u? Viscous rubbing does not care how fast the fluid moves — it cares how much neighbouring layers slide past each other. That sliding is exactly the steepness du/dy. A gradient is the right tool because it measures difference between adjacent layers, which is what rubbing feels.
τw is a stress (force per area); to get an actual force on a tile we multiply by the tile's area and point it the way the rub acts. We need the direction the rub points — call it t^ (built carefully in §6) — so we can write the vector force:
Picture: the green sideways arrow in Figure 1 — that's dFτ acting on the tile.
Picture: number of little fluid dots crammed into a box. More dots = bigger ρ = heavier stuff to shove aside.
Where does the 21 come from? It is the kinetic-energy factor. A parcel of fluid of mass m carries energy 21mv2; per unit volume (m/vol=ρ) that is 21ρv2. So q is literally the kinetic energy density of the flow, and the 21 is inherited straight from 21mv2. See Bernoulli's Principle, where q trades against static pressure.
Why we need it: drag comes from throwing fluid mass out of the way, and q is exactly the "how hard moving fluid pushes" number. Every drag formula will be q times an area times a shape number.
The formal drag formula slices the patch forces dFp=−pn^dAs (§3) and dFτ=τwt^dAs (§4d) down to their along-flow part. The tool that does the slicing is the dot product — so we earn it now, and we also pin down exactly what t^ is.
Figure 3 — Pressure case. Front normal (n^⋅x^<0), back normal (>0), and top normal (=0). Because pressure carries a minus sign, the front push turns into positive backward drag; the top/bottom patches add zero.
Pressure cases (using −pn^):
Front (Figure 3, left): n^ points partly into the flow, so n^⋅x^<0 — but dFp carries the minus sign, so this contributes positive backward drag. ✓
Back (right): n^ points with the flow, n^⋅x^>0. Equal back-pressure would cancel the front; the low-pressure wake drops it, so it does not cancel — net drag remains.
Top/bottom (n^⊥x^): n^⋅x^=0 — pressure there adds zero drag (pushes purely sideways).
Figure 4 — Shear case. The tangent t^ always runs downstream, so along the sides t^⋅x^≈1 (full rub counts as drag); wrapping over the very front/back nose t^ tilts, so t^⋅x^<1; and on any patch where t^⊥x^ the rub adds zero drag.
Shear cases (using τwt^):
Long flat sides (Figure 4): t^ points straight downstream, so t^⋅x^≈1 — the rub is fully backward, all of it counts as drag. This is why a flat plate edge-on is almost pure skin friction.
Over the curved nose/tail:t^ tilts up or down, so 0<t^⋅x^<1 — only the downstream component of the rub counts.
A patch whose tangent runs crosswise (t^⊥x^): t^⋅x^=0 — that rub is sideways and adds no drag, exactly mirroring the top/bottom pressure case.
Sign:τw>0 always (it drags the fluid the way the fluid moves), and t^ is taken downstream, so skin-friction drag is always ≥0 — it never pushes you forward.
Picture: walk around the entire body; at each tile take the along-flow slice of dFp and of dFτ, drop it in a bucket; ∮ is the bucket's total. That total is D:
D=form drag∮(−p)(n^⋅x^)dAs+skin friction drag∮τw(t^⋅x^)dAs
That exact integral needs the pressure and shear at every point — usually impossible to know. So we repackage it. Here is the reasoning, step by step:
What sets the scale of the push? The moving fluid's kinetic push, i.e. dynamic pressure q=21ρv2 from §5. Every pressure and shear on the surface is some multiple of q.
What sets the scale of the area? A single reference areaAref — for a bluff body this is the frontal (projected) area, the silhouette the flow "sees." This is a different A from the wetted surface area As of §2 (recall the notation warning there); we keep the subscript ref so they never blur.
Everything else — the messy geometry (how big the wake is, how the pressure really varies) — gets swept into one dimensionless number, the drag coefficientCD.
Multiplying the push-scale by the area-scale by the shape-number:
Why this is the same physics as the integral: the two ∮ terms are both of the form "(a pressure) × (an area) × (a geometric factor)." Since every pressure on the surface scales with the dynamic pressure q, and every area scales with the reference area Aref, the whole messy sum must collapse to qAref×(some pure number) — and that leftover pure number, carrying all the shape and wake information we couldn't compute by hand, is exactly CD. So the packaged law is not a new law: it is the surface sum D=∮(…) with every unknown quarantined into a single experimentally measured coefficient. That is the whole trick — hide the hard geometry in CD, keep the easy physics (ρ, v, Aref) explicit.
Pointers you will meet next: the Reynolds Number decides how the flow behaves (and whether CD stays roughly constant), and drag feeds Terminal Velocity when it balances weight.