BH App Archive
Quick start – There are two big concersn with coffee roasting. One is an under-roasted centre of the bean with an over-roasted outside. Another is the energy required to do the roasting. The science of both effects is surprisingly non-intuitive and you really have to play with the few factors involved to see how to get the most even roast in the shortest time with the least expenditure of energy.
There is a difference
between “heat” and “temperature”. You can hold some red-hot materials such as the old Space Shuttle tiles in your bare hand, while your hand can be burned touching a piece of metal many hundreds of degrees colder. The difference is that there is little heat flow from the low-conductivity tile and a large flow from the high-conductivity metal.
Now apply this to coffee roasting. If each bean is in contact with 200° metal, then heat flows very quickly to the surface fo the bean, with the risk of over-roasting it, while the inside is relatively cool. If each bean is in contact with relatively still, hot air, heat flows in slowly, so the centre has a chance to catch up in temperature with the outside.
The confusion between heat and temperature is revealed in the two classic schools of roasting: LTLT versus HTST, Low Temperature Long Time versus High Temperature Short Time. If the basic heat transfer process (e.g. in a fluidized bed) is the same, then there is no difference in the balance of core versus surface. The real difference (using “Flow” as shorthand for “Heat-flow”) is between High Flow Big Differential and Low Flow Small Differential, i.e. from a roaster where the beans are in contact with a lot of metal and one where there is a poor airflow. The ideal is one with a good airflow and zero metal contact so that every bean has Optimised Flow with Small Differential.
To model this you can set the relative HTCF – Heat Transfer Coefficient, which is how fast heat flows into the surface and the relative k, thermal conductivity of the bean, which is one of the two factors that govern how fast the heat flows within the bean. A low HTCF and high k means a slower, more even roast, a high HTCF and low k means a greater risk of over-roasting the outside.
Δt the Differential Time – What we should worry about is the difference spent at “roasting” temperature between the core and the surface. The app lets you set “Above T” which you might consider as a temperature above which roasting takes place. It then calculates tout, tin, Δt and % which are the time spent above that temperature at the outside, on the inside, the difference (Δ) between those times, and the % difference (Δt/tout).
My claim that things are not intuitive becomes clear when you change values and worry about Δt. The reason things are tricky is the second factor that governs heat flow within the bean. Whatever the HTCF and k, the heat flow is also driven by the temperature gradient from outside to inside. If for some reason the outside gets hotter rather faster then there is a bigger temperature gradient and a faster catch-up.
Pre-heating – It seemed obvious to me that if you could pre-heat the beans via a relatively slow, energy efficient technique (lots of metal-to-bean contact) to a temperature high enough to be helpful and low enough to have no significant effect on the bean, you’d get a more efficient roasting process (less need for expensive fast-moving air) and a more even roasting. I was right about the former, but wrong about the latter – pre-heating makes no difference to Δt!
Even so, my reading of the literature suggests that up to about 140° nothing much happens to the beans over a, say, 15min timescale. So you can have a nice, high metal surface contact heater taking the beans efficiently up to 140° with no air flow to take away any vital ingredients. These pre-heated beans can then use the expensive high-velocity air in the fluidized bed to get to roasting temperature.
If the inside of the roaster is covered with a low thermal conductivity material (Space Shuttle tiles!) then no bean ever gets into dangerous fast heat transfer to risk over-roasting.
Cooling – Cooling in air requires a similar high velocity airflow, which is energy hungry. Good contact with metal (i.e. something like a cold version of the pre-heating box) is the fastest way to bring the outside temperatures down. The inside gets a few extra seconds at higher temperatures to correct for some of the under-roasting – this is the equivalent of letting a roast joint of meat “stand” for 20min to allow the inside to carry on cooking without over-cooking the outside.
Wild ideas – One question is how much of roasting is chemical reactions (such as the Maillard reaction) among the bean chemicals and how much is “oxidation”, i.e. a sort of burning from the oxygen. My view is that roasting in an inert atomsphere would allow the pure bean reactions to take place, offering a less violent, and probably more interesting set of flavour reactions. If the inert atmosphere was CO2 then you’d even keep plenty of CO2 in the bean to enhance the espresso crema!
Because flowing an inerting gas is expensive, why not have a self-contained system – a sealed roasting system at high pressure? Again, if the gas is CO2 then your crema will be wonderful. More seriously, at higher pressure, the density of the gas is higher so its heat transfer coefficient is higher – so heat will more efficiently get to the bean. Although high pressure systems are expensive, there will be offsets in terms of energy use because the internal energy is much better utilized. I am reliably informed that this is a really stupid idea – but it’s fun to discuss it!