Developing a Kenya roast profile using "Area Under Curve"

(See also: http://kostverlorenvaart.blogspot.nl/2016/11/the-area-under-curve.html)

Arnoud from Hoofdkwartier kindly shared the opportunity to acquire some Kenya, Kianyangi washing station arabica coffee from Dorman. A parcel with two 15kg vacuum blocks were delivered by airmail, one for Arnoud, one for me. Certainly not cheap but very promising. The smell of the greens is delightful so I wanted to take care not to ruin the product and I even postponed giving these a shot for a week or so.

Before this batch I roasted 15kg's of Brazilian beans and they were much cheaper but very tricky as well. They hardly made a sound so there was no way to know when First Crack was happening. Later on there was the audible rustle of Second Crack and around Dry End there was an amazing amount of smoke coming out, but otherwise the cryptic beans were hard to master. When I'd about given up on them they proved to be great to blend with some other roasts though and now I'm enjoying the last bit of them. Other bags have been sent off to friends and family and they seem to do very well on basic small Gaggia-style home espresso units.

Back to the Kenians. As always, I try to take my time to find a roast profile.

Since I don't do cupping, I need to wait a few days or a week to really be able to evaluate how a roast is doing for espresso on my Londinium lever machine. The single doses are ground between the 120mm flat burrs of the R120 grinder built by Compak and extractions are timed using the Acaia Lunar scale.

I first measured the moisture using the Wile device: 10.5%

Then I went through my earlier roasts of beans with a similar moisture value, selecting one that I logged as successful, keeping that profile near by on the desk as I started the roast.

The first batch of just 1kg went a bit faster than I had planned, just 10.5 minutes, a development time of 2:46 making up 26% of total roast time and an Area Under Curve of 203ºC*min.

For a clarification of the Area Under Curve concept, please read the blog by Marko Luther on the Artisan blog pages.

In this case, I let Artisan start the calculation of AUC from the Dry End moment onwards, configured like this, in the Statistics menu:

Start AUC from Dry End onwards,
present me with an on screen Guideline where to end the roast
using the Background roast as a guide, showing the
AUC count / countdown in an LCD above the roast

The beans had an okay color but their aftertaste was a bit dry. They looked good enough but tasted a bit thin, evoking skinny marathon runners who fly by, raising dust. Still they combined well with the more full bodied Brazilians of thick crema looks and after some more resting together they have made me many sweet cups with a voluptuous mouthfeel.

When a company of roasters visited for an evening workshop of sharing a meal, ideas and views together, I showed them the first Kenya batch and we tasted the results together (that's before the blending). Two young roastmasters had a plenty suggestions for the next roast and they easily took control of the Coffee-Tech FZ-94 roaster which they had had no prior experience with and coming from another roast logging software package they also found the transition to Artisan very intuitive so I could just observe how they were handling this roast expertly:

Their roast profile was a full minute longer, mostly taking a more relaxed stride in the earlier stages of the roast and a similar development time. I did not write down the color but the beans do very well, mixed in with the batch that was already in my E8 hopper.

What I learned from these young roastmasters is to keep the airflow relatively low during the first stages, saving the potential increase for the development phase. AUC = Area Under Curve went from 203 in batch 1 to 217 in this Batch 2.

For the third batch, I felt confident that using the approach of a relatively low airflow, I could extend the roast time, getting more energy into the beans before they got to the 'drop' color. I pre-heated the drum a little less than before, to 145ºC and I increased the load of green beans to 1.25kg to slow down the roast curve:

This was roasted on January 6th and when tasted on the 12th, I noted it was the best so far but still a hint of a dry aftertaste told me the roast could take more time.  Roast time went from 11:40 to 13:37 so that's nearly two full minutes extra, covering an AUC of 299ºC*min for a roast color of Tonino #93.

Batch 4 was a slightly longer roast again but aiming for the same AUC, so I carefully watched the vertical line that moved along on the Artisan screen suggesting me where to end the roast if I aimed for the same AUC as the previous profile loaded in the background, and I also kept an eye on the development % displayed in the Artisan screen, since I wanted to be around 25% which is mostly successful in my experience.

The longer roast time primarily benefited the middle part of the roast, going from 4:10 in batch 3 to 4:50 in this 4th batch:

With 297 ºC*min the AUC turned out to be practically identical to the target of 299 and the Tonino value of #95 is also fairly spot on the #93 target.

A first (admittedly premature) espresso from this roast (extracted one day after the roast on beans that were ground to do a Tonino color measurement) gave me a creamy spicy shot. Tasting darker than one would expect from Tonino #95, more like a Tonino #85 roast. 

Some modern baristas might argue this specialty bean deserves a lighter roast as most fruity acidity is traded for spicy full bodied sweetness, whereas more traditional friends coming over for coffee would still say they find the coffee 'a little bit sour.' So it depends on how much specialty coffee one has learned to appreciate.

For Batch #5 I followed the profile of Batch #4, had an AUC=307 C*min and a color of Tonino #97, so both fairly close:

In developing this roast profile, the roast management and evaluation tools in Artisan and especially the AUC value is assisting me greatly.

I hope others will also share these adventures of developing roast profiles.

Please note that Artisan is not subscription based and no purchase fee is required. Anyone can download, install and use this software for free, but it is based on voluntary donations from users who believe that programs like these are made by people who need to be sustained. If Artisan supports your roasting, you should support Artisan. Feel free to do so at their website:
https://github.com/artisan-roaster-scope/artisan#donations

PS Bob, a friend who has a Behmor roaster a Tidaka CB1 grinder and a Strega espresso machine, received a bag of my Kenya roasts. He found it had a sharp taste when hot which is gone when he extracts into a cool cup. "Nice body, delicious soft taste with enough fruitiness to make it interesting. On the Behmor there is a distinct 'roasty' taste which is less successful."

Pictures by Bob:

Picture by Bob

picture by Bob

Left: beans from Behmor                    Right: beans from FZ-94