There have been numerous discussions on the home roasting forum of HB regarding Scott Rao's commandment that "the bean temperature progression shalt always decelerate" - otherwise known as a declining rate of rise (RoR). I've read many of them and found some of them interesting or helpful, but too often they seem to involve an obsessive fixation on whether a roast profile curve satisfies his prescription or not and, if not, what the curve needs to look like. Rao based his rule on his own professional roasting and cupping experience, in his book he says "I've had the opportunity to cup and view the roast data for each of more than 20,000 batches roasted on a variety of machines by various methods." Others in the forums have said that after implementing his approach they have had better results. So I am going to assume that there is definitely something to what Rao is saying (he's clearly got way more experience than I do), which I guess makes the curve fixation somewhat understandable.

But I always felt that the HB Rao discussions left out some important considerations. First, what's really going on to produce the dreaded flatline, crash, and rebound, and, second, why does avoiding this (ie, having a steadily declining RoR) produce better coffee? Rao's answer has always been that roasts without the steadily declining RoR taste baked. To me, baked coffee tastes flat and bland. It's a way of saying that it lacks the vitality that the right kind of acidity produces in a good cup of coffee - a fruity acidity that makes the coffee taste juicy and wet, not sour or astringent.

With that said I'd like to take a shot at answering the two questions posed above, but first a disclaimer, I have no technical or scientific experience in this area. I'm just an interested hobbyist roaster who's read a fair amount about roasting and spent some time thinking about it as well. To answer the question about what produces the flatline, crash, and rebound we need to think about the thermodynamics of coffee roasting. As most roasters know the process starts out endothermic - which means that the beans are absorbing heat from the environment of a hot roaster. This goes on for about 7 or 8 minutes until the beans are approaching first crack. But shortly before first crack something very important happens, sufficient energy has been added to the beans that the reactions within the beans become exothermic (ie, they generate their own heat, like the way a lit match generates heat). This is why you see the flatline before first crack. In fact, with some beans and if you approach first crack with enough energy you'll actually see an uptick in the RoR before first crack. All the while, as the beans approach first crack the pressure inside is increasing as the moisture within the bean cores is turned into steam. First crack itself is nothing other than the pressure from the steam inside the beans shattering the cellular structure and literally cracking them open (hence the dramatic increase in size). As the rate of first crack accelerates the temperature recorded by the bean probe may show a sudden drop or crash. Most discussions on HB that I've seen have described this as a cooling effect from flash evaporation, but I think this is mistaken. The reason the probe reads a decrease in temperature is due to what's known as the Joule-Thomson effect. This effect describes the relationship between changes in pressure and temperature and is responsible for modern refrigeration systems. Think of how in cold weather you forcefully blow on your hands to warm them up - the increase in pressure results in an increase in temperature. Similarly, a sudden decrease in pressure results in a drop in temperature - eg, if you are a cyclist who has used CO2 cartridges to inflate a tire you've noticed the ice that forms on them immediately after use. This is what is happening in your roaster. The intense pressure within the beans (25 atm or over 350 psi!) is suddenly being released and this drop in pressure results in a cooling effect within the air in the surrounding drum. Once first crack is over the beans continue with their exothermic reactions and now you see a sudden increase in the RoR or the dreaded rebound or "flick". One important thing to point out now is that there are NOT two exothermic phases in coffee roasting - sometimes you see people say it's exothermic before first crack and then again before second crack. That's wrong. Once the beans are hot enough they are exothermic and producing their own heat, which happens slightly before first crack, continues through the crash, and into the flick. But one thing that is happening is that the exothermic process is increasing as the temperature of the beans increases, this is why the bump before first crack is typically gentler than the one that comes before second crack. So I think that's a decent start to an answer to the first question of what's leading to the flatline, crash, and rebound (or flick).

The second question, regarding why coffee roasted with a declining RoR tastes better is a much more difficult one to answer and involves some definite speculation on my part, but I think the answer to the first helps to provide a reasonable hypothesis. Generating a more consistent declining RoR means eliminating the flatline or bump before first crack as well as the rebound or flick before second crack. First let's start with how you would actually do that. I see lots of people focusing on avoiding the crash by adding heat to the roaster in anticipation of the crash. I think this is wrong. You should focus on the pre-first crack bump and pre-second crack flick. To do this you need to back off on the gas in anticipation of each event. For example, first crack typically starts on my roaster around 390F and the flatline or bump might happen somewhere around 370-380F. But changing the temperature in a roaster is like changing course in large ship, you need to initiate the change well before you want it to happen. So this involves dialing down the gas somewhere in the 350-360F range (about a minute before you might see the bump), how much requires experimentation as each bean and roaster is different. Similarly, I back off on the gas around the peak of first crack to avoid the impending flick. Doing these two things alone will give you a much better looking curve in terms of Rao's principles. You will still have something of a drop in temperature due to the Joule-Thomson effect, but it will actually be less. In fact, the phenomenon itself is unavoidable as it's basic physics, but you should be trying to use the cooling to your advantage as it actually helps to control the exothermic acceleration of the beans occurring at this point in the roast. If you try to add heat to avoid the crash you are just setting yourself up for a much bigger problem with the flick - a runaway exothermic reaction. So why does coffee roasted without the pre-first crack bump and pre-second crack flick taste better? First of all, chemical reactions occur due to temperature and pressure. The increased pressure inside the bean prior to first crack is critical to advancing those reactions and converting nasty acids into tasty acids. If you rush too quickly through the period when the beans are under a high internal pressure and into first crack you eliminate a lot of these reactions. You need enough momentum to get into first crack but you want to go into it slowly. Rob Hoos has a really good discussion of this starting on page 53 in his book. Second, if you go into first crack a little more slowly and spread it out more it will have the benefit of easing the crash that occurs due to the Joule-Thomson effect. The reasoning behind why the flick is bad is a little different. By this point in the roast you should have already essentially created all of the aromatic flavor compounds (remember that light roast coffees are at their aromatic peak). The flick involves a rapid increase in the RoR due to the exothermic reactions in beans and you run the risk of damaging these delicate compounds with an overly vigorous series of exothermic reactions (but keep in mind that these delicate flavors will be gone in any case if you go very far into second crack).

So far that's my read on coffee roasting thermodynamics. No doubt a lot of what I said above, particularly in response to question 2, is very speculative, but I think an understanding of the thermodynamics of first crack helps to tie together Rao's observation and some reasonable tactical approaches to roasting coffee that have provided me with my best results yet. I hope this discussion was helpful and I'm curious what others think.

Interesting read. I agree that it is wrong to think of the dip as due to an evaporative phase transition, as the water inside the beans should have turned to steam well before FC. One question though. I get why loss of pressure inside the beans themselves might induce a loss in internal temperature (or perhaps just a decline in the rate of temperature increase -- I'm not sure which would be the case inside the bean in a real roasting environment). But why would the air pressure surrounding the beans go down? The bean surface shouldn't change temperature too much (it was not under pressure at any point) and doesn't the loss of internal gasses into the roasting environment slightly raise the roasting environment air pressure, rather than lower it?

My own theory (which might work in concert with yours, I'm not sure) is that the internal bean temp is lower than the exterior temp, and so as the internal steam escapes it briefly exerts a cooling effect on the surrounding air.

Has anyone ever been able to measure the internal bean temperature while roasting? Seems like, while insanely difficult, not outside the realm of reality? Just a random thought that would help to shed a lot of light on this topic.

Clearly you've been thinking about this a lot, Ken, and I commend you for taking the time to formulate concepts, put it all together, put it in words, and throw it all out in the dartboard of this internet forum. It's been decades since my thermodynamics training or use so I'm not speaking from expertise here, but the one thing that I haven't reconciled about the Joule-Thomson relationship with for example the CO2 cylinder analogy, is that it is dealing with the effects of changing from a liquid to a gas where a lot of energy has been put into converting the CO2 into a liquid in the cylinder and its expansion to a gas causes the cooling effect. In the coffee bean I don't think the water is under a lot of pressure prior to roasting, and it is in fact the conversion of the water to a vapor through roasting that creates the internal pressure, and thus the water is already a vapor at the point of release. It seems that since it is now already a vapor, there is no expansion to to a vapor upon its release to cause the cooling.

Here's an interesting explanation of the CO2 cylinder release from a Physical Chemistry professor and others.

[creative nickname] wrote: I agree that it is wrong to think of the dip as due to an evaporative phase transition, as the water inside the beans should have turned to steam well before FC.

This has bothered me too.

texmachina wrote:Has anyone ever been able to measure the internal bean temperature while roasting? Seems like, while insanely difficult, not outside the realm of reality? Just a random thought that would help to shed a lot of light on this topic.

Yeah, its been done by thermoprobe and mathematical modelling. But they don't use drum roasters - and this makes all the difference in measurement, because they don't see a crash. The results aren't usually presented in a way that's useful for home/commercial roasters and they're using air or lab style/built roasters.

Yes, I've seen this study too. Jacob, let me know if it is important to you that I dig it up the reference information to post here, unless Chris already has it handy?

Sorry I don't have any references handy - there are so many studies on it now and yet they don't report their data in a way we can use.

Jim, that would be awesome, but don't spend too much time on it if it becomes illusive.

I was thinking more along the lines of ultrasound; something like this:
https://ieeexplore.ieee.org/document/4258123/

I assume the greatest challenge in a method like that is the dynamic nature of the beans themselves while roasting, but man how cool would it be to be able to render a visualization of the thermodynamics of a coffee bean, responding to different stimuli (conductive, convective, etc...).

[creative nickname] wrote:I agree that it is wrong to think of the dip as due to an evaporative phase transition, as the water inside the beans should have turned to steam well before FC. One question though. I get why loss of pressure inside the beans themselves might induce a loss in internal temperature (or perhaps just a decline in the rate of temperature increase -- I'm not sure which would be the case inside the bean in a real roasting environment). But why would the air pressure surrounding the beans go down? The bean surface shouldn't change temperature too much (it was not under pressure at any point) and doesn't the loss of internal gasses into the roasting environment slightly raise the roasting environment air pressure, rather than lower it?

Yes, that's what I thought about the water already being converted to steam before FC, which means it can't be evaporation that causes the observed cooling effect. So then what causes the temperature decline? And that's when I thought it must be the pressure change. Although maybe pressure isn't the right word, maybe expansion is better, but not sure I fully understand the physics of it. Anyway, inside the bean is a closed space and once the steam escapes the bean it is in an open or uncontained space so the impact on the pressure inside the roaster would be negligible. Check out this video.

The passage below is Rob Hoos take on the potential role of bean pressure in roasting:

Rob Hoos wrote:An offhand comment by Illy and Viani in Espresso Coffee: The Science of Quality got me thinking about the importance of building up pressure in coffee: ". . . a build-up of pressure within the bean is important for the generation of sufficient aroma." I also saw a note in Coffee Flavor Chemistry saying; "Kaufman (1951) observed that pressure formed in the beans during roasting is necessary to the proper development of coffee flavor."

I'd add that this may be partly why high grown very dense washed coffees (Kenyans, Ethiopians, etc) generate the most aromatics - the dense bean structure allows for a lot more pressure to develop in the bean during roasting and it may also be critical to whether such beans are properly developed or not. It might also explain the comparatively reduced aromatics of decaf and lower grown coffees.