Guide to Commercial Roasting Technology and Controls

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
User avatar
Team HB

#1: Post by another_jim »

Download and read carefully. This one's from the Norwegian Technical University, and it's the best one I've seen. Seems they don't just drink the most coffee, but also know something about roasting it :wink:
Jim Schulman

User avatar

#2: Post by DJR »

This is an interesting and useful article. Thank you. His conclusion, though is something I don't really understand. Is he saying fast roast = better or fast roast = more profit?

I am now using propane as an external (only) heat source (no attempt to recirculate gasses etc., which is probably not necessary for a one pound roaster, right?). (see: Knife Sharpener Roaster (was Milk Can Roaster) for an account of the evolution of my roaster)

I could get a roast as fast as 7-8 minutes to 2d crack which would definitely give me tipping and some unevenness. I have been slowly zeroing in on a target of about 14 minutes to second crack (stopping short of the actual second crack, quite often and on purpose). at 14 minutes, the beans are beautiful, pretty even, no tipping.

What I can't figure out is, is there an optimum target for time elapsed to end of roast? He doesn't address that. It looks from his graphs that 12 minutes might be better than 14 to keep the volatiles and other compounds from disappearing. Or am I reading his results wrong?

It's easy to control the roast (just increase or reduce heat) and I have been giving it a slow start-- for the first 3 minutes keeping it below 200 F. During this period, the beans are drying out and turning grassy. Should I reduce or increase this initial period?

Then I ramp it to about 350 F and let it coast through first crack. Along the way I can ramp or control the temperature easily. I use an IR thermometer that doesn't measure internal bean temp, but it is repeatable which is of great help. I have a dipstick that allows me to quickly grab beans to see how they are doing and drop them back in. (aside from the knife sharpener, the setup cost me about $80 including the heat source.)

I cool the beans from roast temp to room temp in 15 seconds using a powerful dust collector, which I put a colander on top of. It's amazing -- I can put my hand in the beans 5 seconds after dumping them. The colander is such that the chaff is almost entirely sucked away. Needless to say, this is an outdoor process.

So, here, in summary are my questions: given the above setup, is 14 minutes to 2d crack a good target or should I try shorter? Is it good to allow the beans to gradually come up to temperature for 2-4 minutes before increasing heat? This is probably the most helpful question to answer if anyone can shed light on it for me.

Thanks for any insights regarding the above.


User avatar
another_jim (original poster)
Team HB

#3: Post by another_jim (original poster) »

These are good questions, but you are looking in the wrong place for the answer:.

This is an academic paper, which means no judgments on what is better or worse, or what people ought to do. Moreover, it is addressed to the audiences that read coffee research, that is other food scientists, and the engineers working at billion dollar per annum revenue food corporations who pay for and directly use this research.

What they want to know is not how to roast the best coffee; but how to get it done when the marketing department asks them for a gourmet coffee, or the accounting department asks them to cut their operating costs by 10%.

However, if you read closely, and without expecting your situation to be addressed directly, you can learn a few things.

Your drum appears to be sealed with no airflow; so it doesn't resemble any commercial roaster used in the last 100 years, all of which use inflowing heated air. Instead, it resembles 19th century hand cranked home roasters which typically had 20 to 25 minute roast times. Here's a few things that may apply to roasters of this sort.

The first consideration as not scorching the beans. The paper states that recirculating roasters use higher environmental temperatures (max 500C) than single pass roasters (max 300C), probably because the atmosphere is more reducing (the oxygen levels are far lower), and therefore less scorching. This may also be true of a sealed roaster. However, there is always an upper limit, and the first key to improving roasts is to get some idea of the maximum amount of heat you can use early and late in the roast to avoid charring the beans.

The paper shows heat transfer equations based on air velocity. These interpret as follows: for the same profile, a low air flow roast will start out with hotter air and end cooler, while a high airflow roast will start out cooler and end hotter. This is simply because the higher the airflow, the faster the heat absorption, and the less the heat storage and time lags. In other words, in a no airflow roaster, you can probably sock in the heat major league early on, and then you must really back off as you get close to the first crack.

The paper includes the formula for getting to the maximum drum speed. In a sealed roasters, the higher the drum speed, barring full centrifugal adhesion, the better the heat transfer, and the faster the roast for a given heat input.

Finally, the taste chemistry section: The graphs have the currently used roast profiles for high grade specialty coffees as their benchmark. These are roasted in small batch roasters for roughly 10 to 15 minutes to roughly 205 to 225C, basically from the end of the first crack to the middle of the second. The graphs of chemical compounds' development use the middle of this range, i.e. 12 minutes to 215C, as their reference point.

The paper doesn't say that the chemical levels produced by this reference profile are the ones that will find the most acceptance among the customers of high end specialty coffee; but it makes this tacit assumption. Specialty coffee roasting develops in the market niche above mass marketed coffee, so the incentive is to find the highest quality roast methods without worrying too much about cost. This means that today's specialty roasting techniques are the best people know for getting good tasting coffee. They may not turn out ot be the best with a hundred years more experience, but for now they are the best.

This means that the airflow, heat input, and time/bean temperature profile used in a small specialty coffee shop roasters is the one that at present produces the best tasting coffee. For instance, Green Mountain Coffee has large batch roasters at its Waterbury plant. But for specialty coffee, it runs them in a mode that emulates the thermal properties and profiles of small shop roasters fairly closely, rather than in the more efficient roasting modes it uses for its K-Cup coffees.

It remains an open question within the specialty coffee world whether 8 minute roasts using high airflow roasters like the Loring or commercial fluid beds are better than 12 minute roasts using lower airflow ones like the Probat. I think the techniques used to produce these charts will settle under what circumstances each will work best.

My personal take, very tentative, based on very limited comparisons of home roasts done both ways, is that the difference is in roast flavors, not acidity or fruit flavors. The 8 minute roasts are analogous to aging in steel or French (uncharred) oak, and the 12 minute ones analogous to aging in American (charred) oak.
Jim Schulman

User avatar
Supporter ♡

#4: Post by Martin »

A tangential point, but this caught my attention:

"Operators of small custom roasters observe the color of bean samples periodically
withdrawn from the roaster (every 15 seconds near the end of a roast) and stop the roaster
when the beans reach the right color."

I realize that color is but one variable in triangulating information, along with temperature/time profile. However, it feeds into my suspicion that color is a more important and under-discussed factor because the tools (mainly eyes) are less efficient and trusted for sharing data. Not quiet scientific enough, along with changes in the radiant temp of the bean mass, character and amount of smoke, etc.

So maybe a question, and I'm not sure that this is the right one:
When roasting are you more inclined to extend time or temp to reach a color you have pre-envisioned; or are you inclined to stick with your profile and (within a range) let the color land where it does?
Heat + Beans = Roast. All the rest is commentary.

User avatar
another_jim (original poster)
Team HB

#5: Post by another_jim (original poster) »

"Color" may be the wrong word. I don't know a lot of people who can rate the luminance of a color within 5 points, even in good light. Pro photographers, who can judge to a 1/3 stop, or about 4 to 8 luminance points at dark to middle gray, might manage to judge match roasting beans to a sample well enough to satisfy a colorimeter; but otherwise it seems a stretch.

I've never seen a pro roaster match the tryer beans to a sample, although some may do this. Instead, roasters look for easil;y detected hallmarks that are specific to the roast level. The first and second crack are the most obvious, but here's a few more I've read about:
-- For the lightest cupping roast, you are looking for the moment the beans to stop being mottled, and to turn to an even color coming out of the first crack.
-- For light commercial roasts, you look for the flat side of the bean (with the split) to become slightly bulged out (light City), or the bean to become completely smooth (darker City).
-- For medium roasts, you may look for the split to start going from pure white to tan (light Full City, but only for WP high grown beans), or to the same color as the bean (dark Full City). For DP beans, you have to rely on the first pops of the second for this roast level.
-- For dark roasts, you may be looking for the beans to bead oil (Vienna) or to be glossy (French).

No warranties. The "bulged flat side" works very nicely for my brew coffees. The split color tell has been hit or miss for me, and I prefer to wait for the moment I sniff the first hints of dry distillate for my espresso roasts. This tends to be about 30 seconds before the first pops of the second.
Jim Schulman

User avatar

#6: Post by Arpi »

Thanks Jim. Those tips really add up.

I always wanted to find a "mechanical" method to tell when to stop the roast (ie. color of a bunsen burner, etc). It is very difficult. For example, the color only indicates the condition of the exterior surface. Just like when cooking a stake at 700F (shear), you need to cook it slowly either before or afterwords (reverse method) in order to cook also the inside. So the exterior color alone cannot be used to judge the end point of the roast if you don't know the past. Maybe if you were to plot the roast, the area under the graph could be used in conjunction with the final color (like the sum of the energy delivered or something like that). But I think that memory and experience are what everybody uses.

Other than smell, is there a "mechanical" way to determine the final sweetness level?

User avatar
Supporter ♡

#7: Post by Martin »

I'm directing this post to home roasters with modest expectations for emulating "professional" equipment, but with high expectations for the quality of their roasted coffee.

I follow a set of useful principles for roasting. These are more or less appropriate for many other home roasters, but they are particularly in tune with my roasting equipment, attitudes, experience, personality, etc. Collectively, my roasting methods are at the opposite end of a continuum of commercial and small-batch professional artisan roasting. I try to learn from, but not emulate, that opposite end. Even so, I benefit from "tips" that others derive from high-level research and practices.

I begin roasts with these few principles (there are others, I suppose) suited for heatgun and 96 oz dogbowl, as a baseline, and then adjust, experiment, or correct. Different batch sizes and bowl sizes are variables that matter quite a lot. I suspect that the same is true for high-grade roasters, but each one, I imagine, has a sweet-spot batch size that should be mastered. 90% of my roasts are 10oz in a 96oz bowl. HG/DB roasting allows for enormous flexibility, all the more reason for keeping it simple.

I do not know (and I'm not sure I want to know :D ) whether those with the most sophisticated equipment and most precise measurement and tracking capability produce roasts that are much better than mine. Overall, given heat and agitated well, the beans know what to do.

1. Steady application of heat, 1st c. at 6-7 minutes. Beans heat with optimal evenness.
2. Ideally, 1st c. extends not much more than 2+ minutes, but some varieties don't cooperate.
3. If unfamiliar with the variety, I roast until a few pops into 2nd--usually 10 minutes. Less, if close to city roast is recommended.
4. Beans roasted in an open environment (such as a bowl) give immediate useful feedback: Color; Heat; Smell; and Smoke. These might not be precisely understood at any instant, but pay attention to trends and changes. These sensory inputs are identifiable and useful in very tiny incremental changes.
5. As nearly as I can figure, beans rarely "require" a roast longer than 10-12 minutes. Anything slower than that is a function of the roaster's capacity for a desired batch size. Besides, longer roasts get boring.
Heat + Beans = Roast. All the rest is commentary.

User avatar

#8: Post by DJR »

My experience in heat gun roasting is similar to yours, Martin. I had very good results and after more than 200 batches I really got the feel of it. One can stop thinking about this or that and one just knows when to stop, how much heat to apply etc. In fact, I'd argue that for small batches this is the most accurately controllable method. Accurate in terms of getting the results you want by instantly changing heat and agitation. (accurately repeatable is another matter).

In the 200 plus batches I roasted I never had a ruined roast due to burning. I did ruin a few by cooling with water. They came out sour even though they looked perfect.

I changed to my new roaster (which I have been using and modifying gradually) for a few reasons. However, if I hadn't done the heat gun roasting, I'm sure I'd have no success with the the "knife sharpener roaster".

My reasons for changing:

1. Batch size -- I wanted to do a pound or so at a time. Heat gun is possible, but tricky. Knife sharpener roaster does 1.5 pounds easily.

2. Boredom -- I got tired of stirring and holding the heatgun.

3. Energy source -- wanted to switch to propane.

4. Cooling -- I have an extremely fast cooler made from a woodshop dust collector. Without a way to cool the beans almost immediately, 1.5 pound roasts are counter productive because they take way to long to cool, even with a house fan. It also removes the chaff.

5. Tool justification -- needed to further justify existing metal working equipment.


User avatar

#9: Post by Arpi »

Arpi wrote:I always wanted to find a "mechanical" method to tell when to stop the roast
Stay tuned for next weekend folks. I think I found the way to happiness, a way to tell the sourness, sweetness, saltiness, and bitterness, at any time of the last final phase of roasting. That means I can stop the roast at the best moment. No especial nose, no especial equipment, and no experience needed. I can't wait! This will be a systematic approach for me to stop the roast at the best time for whatever I am looking for.

Here is my master plan!

I am planning to get two of these puppies: ... ombrowse=0

They have 24 positions. Each column is a minute and each hole is 15 seconds. Total 6 mins of recording time (6 columns = 6 mins).

The day before I do this sampling test.

I do a 200 gram roast. During the final phase of the roast I sample every 15 seconds 4 grams (4 beans?) and place the beans in each hole. Get 2 more extra beans and place them in the second tray.

Then I'll grind each hole in a glass and fill the holes of the tray with ground coffee (replace beans by ground coffee in each location)

Pour 200F and sample the whole tray! Each hole is 15 seconds from 1C or a specific BT

This idea came from this other grinder thread where I learned how to do minimalistic cuppings.

Grinder adjustments for drip

Haven't done it yet but this technique may be my roasting glasses in my darkness of ignorance. Sadly, if anybody wants to try it, they will need a consistent roaster with a sampler, and a batch grinder.

I forgot, the two extra beans stay in the second tray unground (whole). These would help identifying the right color in conjunction with the time of the final phase (next roast).


User avatar

#10: Post by Dieter01 »