The Area Under the Curve

(also see http://kostverlorenvaart.blogspot.nl/2017/01/developing-kenya-roast-profile-using.html )

Marko Luther is a very prolific coffee innovator with products like the Tonino roast color meter which compared to the competition is at least as precise for a fraction of the price.

Still, he may be best known for the free-for-all roast logger Artisan which he has been upgrading and updating intensely over the past years.

Last week Marko posted two new blogs which are both very interesting.

First he shared a very thorough exposé about PID control, explaining the basics of PID and detailing how Artisan now offers PID control even for those users who don't have PID hardware. If you have a device that influences your roast profile and you can regulate it with a slider in Artisan, then you can use Artisan's software PID to "cruise control" your roast.

Then Marko unveiled yet another new feature of Artisan called Area Under Curve, based on a concept that was discussed on the Home-Barista forum.

AUC= Area Under Curve (sketch by Marko Luther)

The idea is that you can define a certain significant part of the curve, for instance starting at 110ºC or right at the Dry End point and calculate how much 'surface area' there is until DROP at the end of the roast.

How is this useful?

Imagine your aim is to repeat a previous roast. You might observe that the Bean Temperature curve has been a little higher than planned early on, then a bit lower and the curve gets less steep in the very last phase. Now do you wait until the BT has reached the same end temperature as in the previous roast, do you try to keep the development phase (the part right after First Crack) timed like the previous roast, do you compromise or do you even revert to other criteria such as checking the visual appearance of the beans, trying to remember how they looked the other day?

Artisan now offers a new assistant. If you have loaded the previous successful roast in the background, you can read the target number for Area Under Curve on your screen and if you select the option, you can see an LCD field with your current AUC number or a countdown to zero of the area that you need to cover to reach the same figure for your current roast. Plus there's a vertical line that shows you at what moment in the near future you would need to end the roast at the current rate.

I tried it out. These are my settings:

I opted to calculate AUC from the Dry End point onwards. The possible advantage is that I could compare the numbers with a fellow roaster, even if he / she has a different machine with Dry End at a different measured probe temperature. I also selected "Background" as the source for my target.

This is the profile I wanted to reproduce:

Guatamala beans, end temp 198ºC in 13:42, 27% development, Tonino # 93 result

As you can see in the above profile, the Rate of Rise soared a bit running up to First Crack and then gently subsided. I used extra fan speed at the start and just before First Crack. One heating element was shut off a while before First Crack.

This is the profile of the subsequent roast with the above roast loaded in the background:

Guatamala beans, end temp 190ºC in 13:27, 25% development, Tonino # 94 result

At the start, Bean Temp was a little lower than planned, then from around Dry End up to First Crack the Bean temp was a little higher all along, indicating that more energy went into the bean mass. One element off after 9 minutes, the second before 14 minutes and the third a little later. With slight increases of Fan speed I aimed to avoid the significant 'bump' in the Rate of Rise and the profile showed only three soft bumps up to First Crack.  After First Crack the Bean Temp stayed a bit below the plan but the countdown LCD and the vertical line suggested I stop the roast at about the same time as in the original roast.

I followed the suggestion, ending maybe a fraction sooner as soon as I saw "25%" development time and the result was a roast color of Tonino #94 which is practically identical to the #93 I was aiming for.

I will play around more with these new features.

Thanks Marko for providing these!

PS 3 december 2016: started with a hotter drum and lower airflow, gradually increasing airflow to help decrease Rate of Rise. Must keep that up through FC next time. Roast went a little faster than the background roast, AUC was a little less but watching the AUC, the development-% and the vertical help-line suggesting when to end the roast was great support in on-the-spot decision-making. Got a Tonino #91 which is still close to the target. A lower RoR in the end could have yielded a more perfect match.

PS 4 dec 2016 -- the profile below has the profile from above as a background:

Much similar, just a tiny bit less aggressive and a nicer declining RoR after FC. I started with a drum pre-heated at 170ºC, so about 10ºC less hot than yesterday. I now knew to keep gradually increasing the airflow (brown dots, light blue line above) and not hesitate to keep doing that after FC. I switched off the elements one by one when the voice alarm of Artisan instructed / reminded me to do so (the grey dots on the display above). The result was a fractionally lighter bean at Tonino # 92.

An example of the vertical help-line suggesting that at the current AUC rate the roast could end a little before the 11 minute mark:

In the end I switched off the last element a bit too early and there was not enough heat to pull up the BT in the end. I quit the roast when RoR went to zero. With 24% development the beans came out on the light side, Tonino # 111 and we'll see how that works out in the cup.

Next roast the BT ended at 192ºC as planned but half a minute sooner, compensated for by the higher temeratures in the last phase with a gently decreasing RoR, yielding a Tonino value of #99. I've already had #99 batches from these beans that tasted very good, so expectations are high:

PS 25 december 2016. In the background of the profile below is Batch #43 of my Guatemala beans. It was okay but tasted a little "too light" and measured Tonino # 103 so I aimed for a longer roast time and a slightly higher end temp. That worked but early on in the roast the curve was lower so the AUC of the two ended up close together: AUC=537C*min for Batch #43 and AUC=550C*min for this Batch #46. Tonino value was also almost the same: T#105 even.

Will be interesting to see if the somewhat dry aftertaste that indicated "too light" for me is still there or if the longer overall roast time and slightly longer development time if this roast works well: