Coffee roasters and heat transfer

Three routes for heat, and where each matters

Sourced

A doctoral thesis on roasting sets out the three routes and adds that roasters generally use all of them, in proportions that may differ greatly.

A model of the routes, not a measurement of any machine. CoffeeHQ read no source that measures the share of each route in a particular roaster, and gives none.

How the beans are kept moving

Sourced

Horizontal rotating drum

The beans tumble inside a turning cylinder. The thesis names it first among the systems most commonly used, and three of the home machines whose manuals were read for this site are drum roasters.

Fixed drum or bowl with mixing elements

The vessel stays still, or a bowl turns on a vertical axis, and paddles do the mixing. The thesis lists these among the systems most commonly found in industry.

Fluidised bed

A column of hot air lifts and circulates the beans, so the air is both the heat source and the mixer. Small home air roasters apply the same principle.

Batch or continuous

A separate distinction. Batch roasters take one load at a time and give more flexibility in how a roast is run; continuous roasters are built for large hourly output.

Why mixing is the real design problem

Sourced

The thesis states the task of all these mechanisms in one line: to mix the beans well enough to roast them evenly and to prevent scorching. A bean resting against hot metal, or sitting in a still pocket of a batch, takes more heat than its neighbours. Everything else about a roaster's mechanics follows from avoiding that.

It is also why batch size matters to the person using a machine. Makers state the load a machine is built for, and one home drum roaster's manual warns that roasting less than its recommended amount makes over-roasting more likely. A retailer's notes for a home air roaster say the opposite problem appears there: too small a batch traps less hot air and may never reach first crack. Both are instructions for one machine each, and they point the same way, towards the load the roaster was designed around.

Air-to-bean ratio: the number behind drum and air

Sourced

The literature has a measure for what the names only gesture at. The air-to-bean ratio is the weight of hot air used per weight of green coffee over a roast. The thesis cites figures from about one in a typical conventional process to as much as 150 in a fully fluidised bed. On that scale the difference between roasters is one of degree: how much air, relative to coffee, is carrying the heat.

In the thesis's own trials the ratio turned out to matter to taste. Low ratios gave coffee it judged superior in the cup, while very large air streams gave coffee it describes as bland, dull and flat. Its author assumes the reason is that less air strips away less aroma and brings less oxygen into contact with the beans, and concludes that roasters should run with a fairly high share of conductive heat and a low ratio.

That conclusion comes from one laboratory, mostly from one fluidised-bed laboratory roaster and one Colombian coffee, and it should be read as that. A later study that roasted three Ethiopian coffees in a drum roaster, a fluidised-bed roaster and a traditional pan found no roaster best for all three: two coffees scored highest from the drum and one from the fluidised bed. It used one machine of each kind, so it cannot separate the kind of roaster from the particular machine.

Inside a turning drum: what tracking one bean showed

Sourced

What the beans do inside a drum has been watched directly, once. A group at the University of Birmingham, with a coffee company, made a single bean faintly radioactive and followed it through a small drum roaster with a medical scanner, a technique that sees through the steel. The roaster was cold. To stand in for a roast they ran three coffees, green, part-roasted and roasted, at three batch sizes and three drum speeds.

Every run showed two regions. Most of the beans sat in a dense bed at the bottom, moving slowly, at up to about half a metre per second. The rest were in flight, lifted by the vanes on the drum wall and thrown across the space above the bed at up to four times that speed. The authors' reading of what this means for heat is the useful part, and it is reasoning from where the beans were, since nothing was being heated. A bean in flight is surrounded by moving air and takes heat from it quickly, for a short time. A bean against the wall takes heat from the metal by contact, for longer. A bean in the middle of the bed gets little from either and waits for its turn.

Drum speed and batch size moved beans between those regions. With green and part-roasted coffee a faster drum put more of the charge into flight, with most of the change coming between the middle and the highest speed, where the motion shifted from beans rolling down the face of the bed to beans being flung clear of it. With roasted coffee the share in flight changed little, but the average speed of a bean nearly doubled. A larger batch did not change the typical speed much but made the beans behave more alike. And the coffee itself changed the picture. Beans lose more than half their density as they roast and swell, so the same weight fills more of the drum; on the group's measurements green coffee rolled and roasted coffee was flung.

From this the authors argue that drum speed and batch size are levers on the balance between heat from the air and heat from the metal, and that a setting that suits green coffee at the start of a roast will not have the same effect on the lighter, bulkier coffee at the end. That is an inference from motion in one small machine. Nobody in the study measured how much heat arrived by which route, and nobody tasted anything. It does replace a guess with a picture: the proportion of contact to air in a drum is not a fixed property of the machine, and depends on how fast it turns and how full it is.

Why a roast curve does not transfer between machines

Sourced

A roast is usually recorded as a line of temperature against time, and it is natural to suppose that the same line on another machine gives the same coffee. Three things read for these pages say otherwise.

The first is what the probe measures. A doctoral thesis on sensors for roasting, written by an engineer who built his own instrumented roaster, puts it plainly: the temperature measured depends on the roaster and on where the sensors are, so that a temperature result is not necessarily transferable from one roaster to another. His own probe sat in the moving coffee and he describes it as reading the temperature of the air at the surface of the beans. It follows, though neither source says it in these words, that a probe in a drum is washed alternately by beans and by air, in proportions that the tracking study shows to depend on drum speed and fill. The guide to reading a roast profile sets out the further delays between a bean's interior and the line on a screen.

The second is the route the heat takes. The tracking group compared its drum with a spouted bed it had studied the same way. In the air roaster the wall is no hotter than the beans and almost all the heat arrives in the jet of air; in the drum, contact with hot metal matters as well. Their conclusion is that coffee roasted in the two will differ even under similar histories of time and temperature. That is stated as an expectation from the physics and was not tested by roasting and tasting in that paper.

The third is the machine's own response. The same thesis found that the plain temperature of the beans told it little about the state of a roast, and got more from the difference between the air going in and the air coming out, which showed the heat taken up while water was evaporating and a repeatable shift at first crack. How a given roaster shows those events depends on its mass, its airflow and its heater. A curve is a record of one machine's instruments responding to one batch. It is a good way of repeating a roast on that machine and a poor way of moving one to another.

The experiment the industry already ran: very fast roasting

Sourced

How fast heat is put in has been tested at industrial scale. Early roasting in conductive equipment was slow, with roasts of more than twenty minutes. Gas firing and fluidised beds allowed far faster heat transfer, and in the 1970s and 1980s there was a trend towards roasts of under ninety seconds, sold as efficient and as giving a lower-density coffee that yielded more in the cup.

The thesis's review records why it was given up. The fast-roasted beans had a different internal structure, expanded more, formed more gas, sweated more oil and ended with slightly more water, and were found to stale faster. Above all, the review says, the coffee had not been optimised for taste, and cites descriptions of it as bitter, burnt and astringent. By 2000, when the thesis was written, roasts of more than four minutes were again the norm, and the profiles each company used were closely kept secrets.

What each kind of machine asks of the person using it

CoffeeHQ explanation

For a home roaster the choice shows up less in the cup than in the routine. A drum machine holds more coffee, takes longer, and gives the operator heat and sometimes airflow and drum speed to adjust. A small air roaster handles less, moves through the stages quickly, and is controlled mainly by fan and heat settings; the retailer's notes for one such machine advise watching closely near the end because it brings coffee through the last stages rapidly.

Both produce smoke and chaff, and every manual read for this site says the same two things about them: vent the smoke and do not leave the machine while it is roasting. Those instructions, and where makers differ, are set out on the home roasting page.

What this page does not say

CoffeeHQ explanation

It does not say which kind of roaster makes better coffee. The one source that reaches a conclusion did so in a single laboratory set-up, and the one comparison of machines found no consistent winner.

It gives no percentage of conduction or convection for any roaster, because no source read measured one, and it does not describe hybrid or recirculating designs beyond naming them, because no document describing how they work was opened. Nor does it compare gas with electric heating or weigh the claims made for infrared burners: the one paper read that mentions the heat source says only that drum walls are heated by gas burners or by electric induction, and no study comparing them was found. The tracking study describes where beans were in one small, unheated drum; it measured no heat.

What to do next

• Reading a roast profile — What the operator's controls change, and what the temperature on the display is a temperature of.
• What happens when coffee roasts — The changes inside the bean that all of this heat is there to bring about.
• Roasting coffee at home — What three makers and a retailer instruct about smoke, chaff and fire for four home roasters.
• Home coffee roasters — The equipment page: whether you need one and what owning one involves.