Why a Moka Pot Needs Almost No Pressure, and Then Makes Plenty

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Because the only part of the job that is fixed is tiny, and the part you control with a grinder is almost all of it. A moka pot needs about 1.47 kilopascals to lift its water, which it gets at 100.4 degrees Celsius, which is to say it gets it from boiling and nothing more. Everything above that is the coffee bed refusing to let water through, and the base will keep getting hotter until it wins. At 110 degrees the base is carrying 42 kilopascals and 40.5 of them are the grind.

That is a system with a demand side set by your grinder, a supply side that rises by 4.1 kilopascals for every extra kelvin, and no feedback anywhere. The valve is not part of the brewing mechanism. It is the only controller in the whole arrangement and it is a last resort.

Lifting the water is nearly free, and that is the surprise

Start with what the pot is actually trying to do. Water has to travel from the reservoir, up the funnel, through the coffee, and out of the chimney in the top chamber. On a 6 cup pot that is a rise of roughly 150 millimetres. The pressure needed to hold up a 150 millimetre column of water is density times gravity times height: 1,000 times 9.81 times 0.15, which is 1,471 pascals, or 1.47 kilopascals.

Put that next to the figure everyone knows. A pump espresso machine works at 9 bar, which is 900 kilopascals. So the lift in a moka pot is 0.16 percent of espresso pressure. One six hundredth. If lifting the water were the whole job, a moka pot would need the faintest simmer and would never raise a sound.

And the temperature that produces 1.47 kilopascals is not interesting either. Water's vapour pressure rises by about 3.56 kilopascals per kelvin at 100 degrees, so 1.47 kilopascals of gauge pressure arrives 0.41 degrees above boiling, at 100.4. The base does not have to get hot. It has to boil, and then it is already done.

So the pressure in your pot is a grind setting

Which means every kilopascal beyond that 1.47 is going into forcing water through a packed bed of ground coffee, and that is the one term in the whole system a cook sets deliberately.

The scaling is the useful part. Flow through a packed bed follows Darcy's law, and the bed's permeability goes as the square of the particle size. So at a fixed flow rate, the pressure the bed demands goes as one over particle size squared. Halve the grind and you do not double the resistance, you quadruple it. That single exponent is why a moka pot is so much more sensitive to grind than it feels like it should be.

Run the consequence on a real grinder. A Tuni G2 conical burr grinder with 70 grind settings publishes a 40 mm conical burr turning at a deliberately low 400 RPM and 70 steps "from fine espresso to coarse French press". Suppose those 70 steps span 200 to 1,000 microns, which is about the right territory. That is 11.6 microns per click. Because resistance goes as one over the square, one click is a change of twice 11.6 over 200, which is 11.6 percent, at the fine end, and twice 11.6 over 1,000, which is 2.3 percent, at the coarse end. The same click is five times more powerful at the espresso end of the dial, and a moka pot sits nearer that end than the French press end. One step is not one step.

Our page on why an espresso shot runs too slowly already says the right thing in words, and deserves the credit rather than a repeat: "Grind is the single biggest lever you have. Finer grounds pack together more tightly and slow the water down." Correct, and the only thing this page adds is the exponent and what it does to a moka pot, where there is no pump to argue with. An espresso machine holds its pressure and lets the flow fall. A moka pot holds nothing: it raises its own temperature until the bed yields.

What happens when the bed does not yield

Here is the mechanism behind the word in the question. The base chamber, with its gasket and filter plate seated, is a closed vessel holding water. The pressure inside a closed vessel holding water is set by its temperature, full stop, and that relationship is exponential. Near 105 degrees it is 4.1 kilopascals per kelvin.

Now put a stove under it. Three hundred grams of water plus about 450 grams of aluminium is 1,661 joules per kelvin of heat capacity. A ring delivering 700 watts into that raises it 0.42 kelvin every second, which at 4.1 kilopascals per kelvin is 1.73 kilopascals of pressure every second. If the bed is blocked, so nothing can leave, starting from a normal 42 kilopascals you reach 150 kilopascals, about one and a half bar, in 62 seconds.

A minute. From normal brewing to the valve, with the knob untouched and nothing in the pot measuring anything. That is the whole answer, and it is why the complaint is always sudden: there is no gradual warning stage in a system whose supply curve is exponential and whose demand is a wall.

So the four things that cause it are the four things that block the bed or raise the temperature. A grind too fine, by a few clicks at the fine end where clicks are worth five times as much. Tamping, which does what tamping is for, which is to raise bed resistance deliberately. A filter plate or basket whose holes are packed with old grounds, which is the same blockage from the other side. And a flame or ring wider than the base, which heats the sides of the chamber above the water line and pours energy into the headspace instead of the water.

Overfilling is the version that actually sprays

There is one more and it matters most of the four, because it changes what the valve discharges rather than when.

The safety valve sits in the wall of the base chamber, above the intended water line. Fill above it and the valve has liquid water on its inside face instead of vapour. When it lifts, hot liquid water at well above 100 degrees goes through a small hole into a kitchen at one atmosphere, and it flashes. Saturated steam at 100 degrees occupies 1.673 cubic metres per kilogram and liquid water occupies 0.001043, so the expansion on flashing is a factor of 1,604. Call it sixteen hundred fold.

That is the difference between a valve venting a hiss of steam, which is the valve working, and a valve throwing a sixteen hundred fold expanding jet of near boiling water and grounds across a worktop, which is what people mean by the pot exploding. Same valve, same pressure, completely different event, decided by whether the water was above or below a hole you can see.

Which is why it is worth saying plainly that our own guide to making coffee with a moka pot never mentions it. That page gets a great deal right. Its instruction to "Use medium heat, not high, since blasting it produces a rushed, bitter, sputtering brew" is exactly correct and the arithmetic above says why. Its instruction to use a medium-fine grind and "not tamp it down, since moka pots work best with loosely filled grounds" is the single most important sentence on it. And its advice to start with hot water in the base, which "shortens time on the heat", is right for a better reason than it gives: less time on the heat is less time spent anywhere near the steep part of that curve. But its only mention of the valve is to "check and clean the safety valve and gasket occasionally", and it never says where the water line goes relative to the valve. That is the gap and this page fills it.

Our own product record gets the valve exactly backwards

A Zulay classic Italian 6 cup moka pot is the pot in question, in cast aluminium, with a safety valve, a heat resistant handle and rinse only cleaning. Its record contains this sentence: A built-in safety valve regulates the pressure inside for a stronger, higher extraction, the mechanism that forces hot water up through the grounds to brew.

Every clause of that is wrong, and in a way worth taking apart because the wrong model leads to the wrong habits. A relief valve does not regulate pressure in normal use, because in normal use it is shut and does nothing at all. It does not produce higher extraction, because it has no effect on the brew until the brew has already failed. And it is emphatically not the mechanism that forces water up: that is the vapour pressure of the heated water in the headspace, pressing on the surface of the remaining liquid. The record's performance text compounds it: the safety valve doing its job means a reliable, consistent brew. The valve doing its job means the pot has stopped brewing and started venting.

The same record's own cons are the honest part, and they are useful: it is "aluminium with a California Prop 65 warning", and it "Makes strong stovetop coffee, not crema-topped espresso". That second one is right and the arithmetic explains it. Crema is a foam of carbon dioxide forced into solution and released on depressurisation, and 900 kilopascals does that where 42 does not. A moka pot is not a low pressure espresso machine. It is a percolator with a vapour pump. Our Zulay moka pot review has the rest of the record.

And it gets the hob backwards too, in eight places

This is the larger defect and it has propagated. That record states, in its tagline, summary, overview, build text, performance text, cons, not_for and ideal_for fields, that the pot is for "electric and induction plates only, not gas or open flame". It also states, repeatedly, that the body is cast aluminium.

Those two statements cannot both be useful, and in fact each one contradicts the other. An induction hob works by inducing eddy currents in a ferromagnetic base. Aluminium is not ferromagnetic and its resistivity is low, so a domestic induction hob at 20 to 100 kilohertz cannot couple to a plain aluminium pot. Most will simply refuse to start and report no pan. Meanwhile gas, the one heat source the record forbids, is the heat source cast aluminium moka pots were designed around and have been used on for ninety years. The two have been swapped.

Read it the other way and it still fails. If the pot carried a bonded ferrous base plate so that it genuinely worked on induction, that plate would make it perfectly suited to gas as well, so the prohibition would still be wrong. There is no version of the specification in which both halves are true. Our own moka pot guide repeated the claim without questioning it, advising readers that an aluminium pot "may specify electric and induction plates only, rather than gas". That sentence has since been corrected to say the opposite, which is that an aluminium pot is made for a gas flame or an electric plate and will not work on induction without a ferrous base. The advice to match the pot to the hob was always sound. The example was not.

And it matters for this page specifically, because the heat source is one of the four things that drive the ramp. A gas flame wider than the base is a genuine hazard for the reason given above, which is that it heats the dry sides of a pressure vessel. That is a flame size instruction, not a fuel prohibition, and the record has turned a real caution into a false one.

The honest limit, and the free technique

The limit is that two of the three numbers in this system are unpublished. Not one moka pot listing we carry states a valve opening pressure, which is the figure the whole safety argument turns on, so the 150 kilopascals in the drawing is an assumption and is labelled as one. None states a chamber height, a basket depth or a bed volume, so the 1.47 kilopascals of lift is a worked geometry. And no grinder we carry publishes a particle size distribution.

That last one is worth being precise about, because it defeated the first version of this page. I tried to compute the bed's pressure drop from the standard packed bed formula and a nominal grind size, and it gives an answer four orders of magnitude too small, because the permeability of a coffee bed is set by its fines, not by the setting on the dial. A manual burr grinder with a CNC stainless burr describes itself as producing "consistent grounds" and "even particle size", and the claim is true relative to a blade grinder and still says nothing about the fines fraction, which is the quantity the physics needs. Its record is honest about what it is for, noting that "finer espresso grinds are harder work" and that it suits "one or two cups at a time", which is also a moka pot sized observation. Our manual burr grinder review and Tuni G2 review have the detail. Worth noting on the Tuni that its two programmed doses do not agree with each other: a single is 12 to 13 grams in 12 seconds, which is 1.04 grams a second, and a double is 18 to 19 grams in 16 seconds, which is 1.16. The same burr at the same speed should have one throughput, and if the single is right the double should be delivering about 16.7 grams.

So the free techniques, and they are the only levers the arithmetic says are large. Go coarser, and go coarser by one click at a time, because at the fine end one click is worth 11.6 percent of your bed resistance. Never tamp, because tamping is the deliberate version of the fault. Fill below the valve and look at the valve while you do it, which converts a sixteen hundred fold flash into a hiss. Keep the flame or ring no wider than the base. Take the pot off at the first gurgle, because the gurgle is the sound of steam reaching the funnel after the liquid has gone, and from that moment the headspace is gas and the 4.1 kilopascals per kelvin applies to a chamber with nothing left to absorb the heat. And once a month, take the valve out and look through it, because every assumption on this page about what happens at 150 kilopascals depends on the one component nobody inspects.

One last thing that is not about safety at all. If the pot brews too fast and tastes thin, the arithmetic says the bed is offering almost no resistance, so the base never gets much above 100.4 degrees and the water passing through is barely above boiling for barely any time. Going finer is the fix, and going finer is also the thing that causes the problem in the question. The whole art of a moka pot is that both complaints are the same dial, and there is no instrument on the pot to tell you where on it you are. Our page on how a hose loses pressure works the same Darcy relation in a case where every number is published, if you want to see the arithmetic done without assumptions.