Where Bernoulli's equation applies
Bernoulli’s equation is not wrong. It is one of the more reliable results in the subject, and nothing below disputes it.
What goes wrong is where it is applied. Almost every misuse in circulation is a correct formula carried into a region where one of its hypotheses fails — and because the formula is right, the error never announces itself.
What it says
Along a single streamline, in a steady, inviscid, incompressible flow:
Pressure, plus the kinetic term, plus the height term, is the same all the way along that streamline. Speed up and the pressure falls; slow down and it rises.
The derivation is the momentum equation integrated along a streamline. Every one of the four conditions in the previous sentence comes from a step in that integration, and dropping any of them drops the result.
The four hypotheses
Steady. The flow must not be changing in time. A pump starting up, a valve closing, a wing beginning to move: none of these qualifies, and the equation acquires an extra term when they do.
Inviscid. No friction. This is nearly always fine, because viscosity is negligible except in a thin layer — but it is emphatically not fine inside that layer.
Incompressible. Density constant. Good for water always and for air below roughly Mach 0.3. Above that the equation needs a compressible form.
Along one streamline. This is the one that gets broken most often, and it is worth its own section.
The streamline condition
The constant in Bernoulli’s equation is a constant along a streamline. Different streamlines are entitled to different constants, and in general they have them.
So comparing the pressure at a point above a wing with the pressure at a point below it — which is exactly what the standard lift argument does — is not licensed by the theorem as usually stated. Those two points are on different streamlines.
There is a rescue, and it is worth knowing because it is what makes the usual argument legitimate: if the flow is irrotational as well, the constant is the same for every streamline, and Bernoulli may be applied between any two points at all.
Flow that started from rest and has not touched a boundary layer is irrotational, which covers the outer flow round a wing. So the usual argument survives — but by an extra hypothesis that is almost never stated, and that fails precisely where the boundary layer has been.
Five places it does not hold
Inside a boundary layer. Friction is not a small correction there; it is the entire physics. The total pressure falls steadily through the layer, which is what drag is. Applying Bernoulli across a boundary layer will confidently produce nonsense.
Through a fan, pump or propeller. Work is being done on the fluid, so total pressure rises. The equation has no term for that, and using it across a propeller disc predicts that nothing happens.
Across a shock wave. Entropy rises through a shock; total pressure drops discontinuously. The static pressure and the speed both change abruptly and the Bernoulli constant does not survive.
In unsteady flow. During a transient — a wing starting up, a water hammer in a pipe — the extra term can dominate everything else.
Between streamlines in a rotational flow. In a wake, downstream of a propeller, or anywhere vorticity has been shed, the constant genuinely differs from one streamline to the next.
The classic misuse
The one worth naming, because it is in more textbooks than any other, is the account of lift that compares the speeds above and below a wing and applies Bernoulli between them.
That comparison is legitimate for the reasons above: the outer flow is irrotational, so the constant is shared. The conclusion is correct, and the pressure difference it predicts is the lift.
The problem is not the Bernoulli step. It is the step before it — the claim that the upper flow is faster because it has further to go, which is false and which the Bernoulli step then dresses in an equation. A wrong premise, processed through a correct theorem, produces a wrong answer with an impeccable appearance.
That is a general hazard and worth naming: an equation cannot repair the assumption fed into it.
Total pressure is the quantity that matters
There is a more useful way to hold all of this, and it makes the failures obvious rather than a list to memorise.
Define the total pressure — sometimes stagnation pressure — as the whole left-hand side:
It is the pressure the fluid would reach if brought smoothly to rest. Bernoulli’s equation is then the statement that total pressure is constant along a streamline, and every failure above becomes a statement about total pressure changing.
Friction reduces it. A fan increases it. A shock destroys some of it. An unsteady flow does not have a well-defined one. Vorticity means it varies from streamline to streamline.
That reframing is what engineers actually use. Ducts, intakes and cooling systems are designed by tracking total pressure from one station to the next and accounting for every loss — and “pressure loss” in a specification always means total pressure, never static.
It also makes drag intelligible: a body’s wake is a region of reduced total pressure, and the deficit, integrated across the wake, is the drag. Measuring drag by traversing a probe behind a body and adding up the missing total pressure is a standard technique, and it works because Bernoulli fails there in a quantifiable way.
A worked misuse
One example carried through, because the abstract version is easy to nod at.
A common demonstration blows air across the top of a sheet of paper and observes it rise. The explanation offered is that the moving air has lower pressure, by Bernoulli, so the higher pressure underneath lifts the sheet.
The conclusion is right and the reasoning is not. The air being blown and the still air underneath are not on the same streamline, and they did not come from the same reservoir — the blown air came from lungs, at raised total pressure. Comparing them by Bernoulli compares two flows with different constants, which the theorem does not license.
The honest account is that the jet is deflected by the sheet, and the reaction to deflecting it is what lifts the paper — a momentum argument, not a Bernoulli one. That the demonstration is usually explained wrongly does not make it a bad demonstration; it makes it a good example of the streamline condition mattering.
What to use instead
Each failure has a replacement, and knowing which is most of the skill.
For unsteady flow, the unsteady Bernoulli equation with its time-derivative term.
For work being added, an energy balance across the device: total pressure out minus total pressure in equals work done per unit volume.
For compressible flow, the compressible form with enthalpy in place of the pressure term, and across a shock the Rankine–Hugoniot relations rather than any Bernoulli at all.
For boundary layers, the boundary layer equations, in which the pressure is imposed from the outer flow and the momentum balance is dominated by the friction Bernoulli neglects.
For rotational outer flow, Crocco’s theorem, which relates the variation of total pressure between streamlines to the vorticity — and quantifies exactly the thing that Bernoulli assumes away.
What the figure shows, and what it does not
The figure is a table rather than a solved flow, which is unusual for this site and is the honest form for this content: the failures are statements about hypotheses, not about a particular field, and drawing them as a computed picture would be dressing up a list.
Where a solved field is informative is in showing regions where the hypotheses visibly fail — which is why the wake figure above appears. The layer next to the surface and the recirculating region behind it are precisely the places a reader should refuse to apply the equation.
Why the equation gets blamed
A closing observation about how this misconception is usually discussed.
There is a genre of correction that says lift is “not caused by Bernoulli” but by Newton’s third law, as though the two were competing mechanisms. They are not. Bernoulli’s equation is a relation between pressure and speed in a flow; Newton’s laws are what the flow obeys. One does not explain lift instead of the other.
The real correction is narrower and less satisfying: the equation is fine, the momentum account is fine, and the specific claim that needs discarding is the one about path lengths. Replacing a wrong mechanism with an argument about which correct framework is more fundamental does not help anybody.
How to check before applying it
Four questions, in the order worth asking, and the whole essay reduces to them.
Is anything changing in time? If so, stop; the steady form does not apply.
Is work being done on the fluid between the two points? A fan, a pump, a propeller, a turbine. If so, stop.
Are the two points connected by a streamline? If yes, proceed. If not, the flow must additionally be irrotational between them — which means asking whether either point is inside a boundary layer, a wake, or downstream of anything that has shed vorticity.
Is the Mach number above about 0.3? If so, the incompressible form is not adequate.
Most misuse fails at the third question, and most of the rest fails at the second. Running the list takes a few seconds and would prevent nearly every misapplication in circulation.
The venturi, and what it does not explain
One more case, because it is the other demonstration everybody meets.
A venturi is a constriction in a pipe. Flow through it speeds up, and by Bernoulli its static pressure falls — which is measurable, reliable, and the basis of the carburettor, the atomiser and a common type of flow meter.
Here every hypothesis holds cleanly. Steady, no work added, all the points on the same streamline running down the axis, and incompressible for a liquid at any speed of interest. It is the textbook case, and it is textbook because it is genuinely simple.
What it is not is an explanation of lift, though it is often pressed into that service. The constriction story for a wing — that the flow above is squeezed between the wing and the air above it, as through a venturi — fails because there is no upper wall. A venturi has two walls and a wing has one, and the difference is exactly what makes the wing’s problem interesting.
So the venturi shows the equation working, and it shows nothing about wings.
One equation, two jobs
A last framing that makes the whole essay easier to hold.
Bernoulli’s equation is doing two quite different jobs in this subject, and confusing them is the root of most trouble.
As a computational tool, it converts a known velocity field into a pressure field. That is how every pressure figure on this site was made: solve for the velocity, apply Bernoulli point by point, contour the result. In that role its hypotheses are checked once, for the whole field, and it is completely reliable.
As an explanatory story, it is asked to say why something happens — why a wing lifts, why a paper sheet rises. In that role it is being asked to supply a mechanism, and it has none. It relates two quantities; it does not cause either.
The first job it does well and the second it cannot do at all. Nearly every misuse in circulation is an attempt at the second, and the fix is not a better version of the equation but a different kind of argument — a momentum balance, or the circulation the flow actually has.
Where the model stops
Bernoulli is not a conservation law in its own right. It is the momentum equation integrated under conditions, and the conditions are the content.
“Irrotational” is doing hidden work in most applications, and it fails wherever vorticity has been generated — which is downstream of every real body.
This essay is about the incompressible form. The compressible version has its own hypotheses and its own misuses.
A table is not a proof. The five rows above are statements of where derivations break, and each deserves its own treatment.
Who found it, and when
Daniel Bernoulli published Hydrodynamica in 1738, and the relation as usually written is closer to the form Leonhard Euler gave a few years later — a common pattern, and one that caused a good deal of friction between them, since Bernoulli’s father Johann also had a claim in the business and behaved badly about it.
The equation predates the concept of a boundary layer by a hundred and sixty years and the concept of vorticity as it is now used by a century. That it is so often applied without its hypotheses is partly because the hypotheses were not available when it entered the curriculum.
Why the hypotheses are usually dropped
It is worth asking why a result so hedged is taught as though it were unconditional, because the answer is not carelessness.
The hypotheses are all satisfied over most of most flows. Air is very nearly inviscid outside a very thin layer; it is very nearly incompressible below a third of the speed of sound; most flows people meet are steady enough; and outer flows that started from rest are irrotational, so even the streamline condition can be relaxed.
So a reader who applies Bernoulli carelessly to a wing, a venturi or a pitot tube will get the right answer nearly every time. The habit is reinforced constantly, and the exceptions cluster in exactly the places a beginner does not look — inside layers, behind bodies, across machines.
That is what makes it a hard misconception to correct. It is not a false belief; it is a true belief held without its conditions, and it fails only when the conditions matter, which is precisely when somebody is doing something that matters.
The pitot tube, done properly
A worked case where everything holds, for contrast with the paper demonstration.
An aircraft’s airspeed indicator is a tube facing into the flow. Air entering it is brought to rest, so the pressure there is the total pressure. A second port, flush with the fuselage where the flow is undisturbed, reads the static pressure. The difference is , from which the speed follows.
Check the four questions. Steady in cruise: yes. No work done between the ports: correct. Same streamline, or at least irrotational outer flow: yes, provided the static port is not in a boundary layer or a separated region — which is exactly why static port placement is a serious design problem and why blocked or badly sited ports have caused accidents.
Below Mach 0.3: for a light aircraft yes; for an airliner no, and the instrument applies a compressibility correction.
So the same equation that is misused on a sheet of paper is, with its conditions checked, the instrument every aircraft measures its speed with.
The shortest correct statement
If one sentence has to survive, it is this: total pressure is conserved along a streamline in a steady, inviscid, incompressible flow, and every failure of Bernoulli’s equation is a place where one of those five words does not apply.
Everything in this essay is that sentence unpacked. Friction breaks “inviscid”. A pump breaks “conserved”. A shock breaks “incompressible”. A transient breaks “steady”. Comparing points in a wake breaks “along a streamline”.
Carrying the sentence rather than the formula is the difference between using the equation and being used by it.
The ladder from here
Nearby: the unsteady form and where it matters; total pressure as the quantity that is actually conserved and lost; Crocco’s theorem; and the compressible form.
Then across to the account of lift that does not need repairing, and to the premise that does.