Espresso pressure: why 9 bar is a habit, not a law of physics

Every espresso machine on a kitchen counter advertises a pressure, and 9 bar is the number that gets treated as the definition of the drink. It is not a constant of nature. It is roughly what the spring in a lever machine delivered in the late 1940s, and the trade has built around it ever since. Physicists at the University of Warsaw put a cafe-grade brewer on instruments and measured what the coffee puck actually does as that pressure changes. The bed of grounds turns out to behave less like a filter and more like something that pushes back.

Where the 9 bar habit came from

The authors trace the convention to the force achievable in the first lever-operated machines, the Gaggia and Faema machines of the late 1940s and early 1950s. A barista pulled a lever, a spring drove a piston, and the pressure that reached the coffee was whatever that spring could produce. Pumps arrived later and were built to reproduce the number rather than to interrogate it.

That is a design history, not a physical argument. Nothing in the chemistry of coffee singles out 9. It is the pressure the hardware happened to make, repeated until it hardened into a specification.

What the instruments saw between 1 and 12 bar

The experiment, published in Physics of Fluids, ran on a cafe-grade machine with a fixed dose of 18.50 g in a 58 mm basket locked into the portafilter, the bed 14 mm thick, ground on a Fiorenzato F64 EVO. The grind was set to the Specialty Coffee Association recommended brewing speed at 9 bar and a 2:1 brew ratio, which means 37 grams of liquid in the cup at 32 seconds. Then the pressure at the basket was varied from about 1 bar up to 12 bar.

Below roughly 5 bar the results are textbook. Flow rises in step with pressure, the way Darcy's law says water should move through a packed bed. Push harder, get more liquid, in something close to a straight line.

Above that the line bends, then stops. The flow saturates. The fastest rise of mass in the cup came at a basket pressure of about 5 bar, not at 9. And around the standard pressure the paper reports something stranger still: an increase in pressure causes a decrease in the average flow, a result earlier work had suggested and these measurements corroborate. Grind size is the other half of that setting, and whether the burrs are conical or flat changes what a given number on the dial means.

Why the coffee puck stops behaving like a filter

A filter is rigid. A bed of grounds is not. Under pressure it compacts, its pores narrow, and the harder the water pushes the tighter the path it has to take. The authors call the bed poroelastic: its own elasticity regulates the flow through it, which is why the pressure and the resistance are not independent knobs.

Dissolution complicates it. As solubles leave the grounds the structure changes while the shot is still running, so resistance is not a fixed property of the puck but something that evolves second by second. X-ray microtomography of pucks before and after brewing showed that reconfiguration directly, cracks and peeling included. The whole of espresso extraction happens inside a material that is changing shape while it works.

Which is why more pressure buys nothing past a point. The extra force goes into squeezing the bed rather than into moving water through it.

What actually lands in the cup

Flow is only half the story. The team split shots into 5 second fractions and read each one on a refractometer. The first and most concentrated drops measured about 25% dissolved solids. That held for roughly 20 seconds, then fell away, reaching near zero at about 60 seconds.

So the front of the shot carries most of what reaches the tongue and the tail carries something much closer to water. Espresso extraction is front-loaded, which is why extraction time is doing more than filling the cup. It is also the argument for weighing what comes out instead of watching the cup fill, and the ratio of coffee to water is the simplest lever most people never touch.

The one thing worth changing tonight

One finding is directly actionable. When the brew was stopped and restarted several times, the flow rate went up without producing any additional dissolution. More liquid, no more coffee in it. The explanation is channelling: the bed delaminates when the pressure comes off, water finds the cracks on the next push, and the rest of the puck goes untouched.

The conclusion is blunt. Any brewing process that involves a temporary pause or a repeated application of pressure may be prone to channelling and should be avoided. That covers pre-infusion improvised by flicking the pump on and off, and it covers stopping a shot to look at it and starting it again.

Watch the gauge next time the pump runs. If the needle sits well above the point where more pressure stops buying more liquid, the machine is working hard for nothing and the bed is absorbing the difference.

And what goes into the basket?

A machine can only work with what is already in the basket, and no amount of pressure rescues stale coffee. Santo Café is honest Mexican coffee, 100% arabica and always freshly roasted, which is the part of the shot no gauge measures.