You might call it a gas pump nozzle.

The reason I cut it in half

is because I want to answer one question.

How do these things know when to turn themselves off?

You hear that click sound?

That's the nozzle turning itself off.

Like, when you go to fill up your car with petrol,

you don't have to worry about it overflowing.

That's because these nozzles switch off automatically

when the tank is full, but how?

My first thought was that there must be some kind

of electronic sensor in there,

but the way these really work is much smarter than that.

There are actually two very clever mechanisms in here.

The first one relates to this tube

you can see that runs to the end of the nozzle,

and the second relates to all these connected levers.

The whole thing's quite compact,

so I built a few different things

to illustrate each part of the mechanism,

to make it clearer.

Let's start with this hole

that's usually located here or here.

It's called the Venturi sensor,

which sounds like it's electronic,

but actually this is entirely fluid mechanical.

It's called the Venturi sensor

because it works on the Venturi effect.

The Venturi effect happens in this part of the nozzle,

but to be able to see what's going on,

I got this made; it's called a Venturi tube,

because it demonstrates the Venturi effect.

It's a wide tube that narrows in the middle,

and then you have this narrow U-bend shaped tube

coming off the restriction in the middle.

There are two other U-bend shaped tubes either side,

but we're ignoring those for now.

Look what happens when I partially fill this U-bend

with water, and then blow through the tube.

You might expect air to be forced into the U-bend,

causing the level of water here to go down.

In fact, the water level here goes up.

That means there must be a reduction in pressure here

in the constriction, and that's sucking the water up.

By the way, the water in the U-bend

is there just so that you can see the change in pressure.

There's no equivalent to that water in the nozzle itself.

But why does the pressure go down?

Well, it's Bernoulli's principle,

but actually, we can explain it quite easily

from first principles.

The air traveling through the constricted part of the pipe

must be traveling faster

than the air in the wider part of the pipe.

That makes intuitive sense.

To get the same amount of mass through a narrower pipe,

the mass must have to travel more quickly,

but you can also think about it

in terms of conservation of energy.

The gas here has kinetic energy,

but it also has potential energy stored as pressure,

a bit like how you can store energy

in a spring by compressing it.

But that total energy, kinetic energy plus potential energy

stored as pressure, needs to be conserved.

That means that when the kinetic energy goes up

in the fast moving fluid in the constricted part

of the tube, the potential energy must go down.

The pressure must go down.

By the way, those two other U-bend shaped tubes

either side of the middle one are there to illustrate

the fact that the pressure goes down

in the wider parts of the tube, as well,

when there's air flowing, but just not to the same degree

as it does in the constricted part of the tube.

The two outer U-shaped pipes have nothing to do

with the discussion we're having here about petrol nozzles;

I just thought I should explain what they were.

But thinking about that central U-shaped tube,

what does that tell us about how the petrol nozzle works?

Well, this pipe in the model is this pipe in the nozzle.

The constriction in the pipe that happens here in the model

actually happens here in the nozzle.

This spring-loaded stopper here opens slightly

under the pressure of the petrol

to reveal a really narrow ring

for the liquid to pass through.

And you can just about see, inside that ring,

there's a tiny hole there.

Actually, there's a number of holes.

The one I just illustrated with the paperclip.

There's also this one, and there's probably one

on the other side as well, but they all lead up to here,

and then into this hole, which comes down through here,

which feeds into this long tube.

This tube that runs to the end of the petrol pump nozzle

is equivalent to this tube in the model.

So, this constriction here creates low pressure in the tube.

There is low pressure here at the end of the nozzle.

That means air is actually drawn in through this tube.

That air simply mixes with the petrol

in this part of the nozzle.

So, how is this used to detect when your petrol tank

or your gas tank is full?

This bit's really clever.

The tube that comes away from the constriction

in the flow of the main pipe is actually forked.

One tube goes off to the end of the nozzle, as we've seen,

but there's actually a second tube that goes off up here.

Because of the way this thing has been cut in half,

it's not that easy to see,

but there is a tube coming off here.

Most of it's been cut away, but it was there.

And that tube leads to this cavity here.

That cavity is sealed by a membrane.

You can see part of the membrane just there,

before it was cut in half.

In other words, before I cut it in half,

this whole chamber was sealed off with a membrane.

So, schematically, it would look like this

with two tubes forking off the restriction.

This now represents the tube

that goes to the end of the nozzle,

and this represents the tube

that goes to the sealed chamber.

So, when I blow through this pipe,

it will reduce the pressure in this tube and this tube.

But look what happens when I put my finger over this tube.

It turns out that this tube

was relieving some of the negative pressure.

And when I put my finger over it, look,

you see a sudden jump in the water level in this tube.

The same thing happens with the petrol nozzle.

This opening is allowing some of that negative pressure

to be relieved by allowing air to flow into the system.

But then, what happens when the level of petrol in your tank

reaches the end of the nozzle?

Well, it covers up that hole.

The tube is now sucking on petrol instead of air.

Petrol is heavier than air, so the tube can't suck as much.

It can't relieve as much of that suction force.

And just like when I put my finger

over one pipe in the model,

the other pipe experiences an increase in suction force.

This is my attempt to put all that together.

So, you've got the main tube.

It has a constriction here.

Here's the tube coming off from the constriction.

Here is where it forks.

One tube goes to the end of the nozzle,

the other goes to this chamber

that's sealed off with a membrane here.

And look, when the liquid in the tank reaches the end

of the Venturi tube,

you see that membrane gets sucked into the chamber.

The membrane moves up only slightly in my model.

That's because I don't know much about fluid dynamics.

I'm sure there's a lot of things I could tune in this model,

like how much is the pipe constricted?

How wide are the Venturi tubes?

Where does the fork happen?

All that sort of stuff.

But for me, it was incredibly satisfying

to see that membrane move at all.

So, when the petrol in your tank

reaches the end of the nozzle, this membrane moves.

And you'll notice that this membrane

is attached to something.

It's attached to this rod here.

And that's really important.

That's how the nozzle actually turns off.

It's quite hard to see what's going on here,

so I built another model.

So, imagine this is a valve that lets petrol through.

I need to push this thing up to open the valve.

In the actual nozzle, this valve is spring loaded.

In this model, I'm representing that fact with a mass.

The mass is pushing back down on the valve

like the spring in the real thing.

And this here is the handle of the pump.

So, hopefully, if I pull this handle up,

it will open the valve.

But look, it doesn't actually work

because we've got a lever happening here.

Pulling the handle up just makes this thing move down.

What I need to do is hold this thing in place

so that this becomes the fulcrum of the lever.

Now, when I pull the handle up, it opens the valve.

Great.

Now, one way I can hold this thing in place

is to put a couple of circles in here.

And then, I put this piece in to hold these circles

in place, to stop them falling into the middle.

And there you go.

That works perfectly.

Now, here's the clever part.

This funny wedge thing is attached to the membrane.

And remember, when petrol reaches that Venturi tube,

it causes the membrane to pop up.

And when it pops up, it pulls this part with it.

And when it does that, those circles are now free

to fall into the middle.

They're no longer jamming that shaft in place,

and this point stops being the fulcrum.

Everything collapses, and the valve shuts.

In three dimensions this is achieved three ball bearings

in these two positions,

and one round the back that you can't see.

So, let's put all those mechanisms together

inside the nozzle itself, and recap.

So, petrol comes in here under pressure,

and it meets this closed valve.

So, you pull on this handle here,

you'll notice it doesn't open the valve.

Instead, this thing moves.

But look, you get to this point here,

and these ball bearings, they get jammed

against the constriction here in the housing.

Of course, because I've cut this thing in half,

that doesn't work, the constriction doesn't work,

and the thing can move when it shouldn't.

So, I'm just gonna use brute force to hold that in place

to illustrate the point.

And look, now... oh, I've lost the ball bearing.

Doesn't matter.

Now, when I put on this handle,

this point acts as a fulcrum, and the lever mechanism

opens the valve, and petrol can flow through.

So, let's imagine the valve is still open;

petrol flows through here.

It looks as if this housing is in the way of the petrol,

but actually there's a gap underneath,

and there's a gap on the top as well.

So, petrol flows around this part of the mechanism,

and then the pressure of the petrol

pushes against this spring-loaded thing here,

which creates a thin circular channel

for the petrol to continue to flow through the nozzle.

Now, because it's a thin channel,