#roast profile

[roastproject]

What do we actually develop when roasting a coffee....

[celtic-roaster-blog]

Each and every time we roast coffee, we generate a roast profile. Our roaster is fitted with a digital temperature probe in the drum and this gives us real-time data as the roast progresses. This is plotted as a graph similar to that above.

We charge the drum to a specific temperature before we drop the beans in. We run at around 10kg batch sizes in a Probat 12kg roaster. Throughout the roast we will manually control the temperature in order to follow a roast profile which we will have determined beforehand. For the first half of the roast the beans slowly dry and they remain green in colour. Approximately 6 minutes in and we start to see a colour change. You can see a time lapse video of this in a previous post. We continue to manage the heat applied to the beans right the way through to the end when we stop the roast by dropping the beans into a cooling drum. The above roast is for Inza, a single origin coffee from Colombia.

[hungryvillage]

Keba Konte explains how he develops roast profiles in his basement "coffee dojo." This is yet more footage from our forthcoming short documentary about him and his various food projects.

[pactcoffee-blog]

Colombian La Joyeria #212

We went on the hunt for a coffee that filled the ‘bright, lively’ category in our coffee portfolio, and we’re thrilled with this one. It’s fairly acidic but really well balanced with a creamy, fruity body. Pop it in a filter, pourover or aeropress and you’re in for a treat!

We tried numerous different roast profiles before settling on this one. It’s fairly, which preserves the coffee’s acidity. Any darker and it was a little too full bodied, which tasted a little imbalanced for Nick’s exacting palate.

This is known as a single picking coffee, which is super-rare. That means that not only is it single origin (country), single estate (grower) or even just a single lot (little bit of the grower’s farm), but that the beans were all picked during the same 48 hour period, from the same spot, when the beans were at their very best. There are only 12 sacks of this coffee in existence, and we managed to snag 3 of them!

The farm itself is called Santa Barbara and is here in Antioquia. It is a model for ethical farming practice (something sorely lacking from most of the mass market growers). All the staff are paid 30% over the minimum wage, and half receive free housing – part of the reason that many have worked here, full time, for a decade. They even assist retiring pickers in purchasing a smallholding when they retire. Hats off to the Echavarria family, and let’s hope more growers follow in their humbling footsteps.

Want to give our Colombian a try?

Existing members log in to your account ->

New customers sign up here ->

[appliedcoffeetechnology]

Developing a Roast Profile on One Roaster for Use on Another

Those of you who have been posts here know that a Diedrich IR-1 has recently been installed at the shop and that I've been working with this lately. There were three potential uses that I had for this machine.

First, it could be a sample roaster. While the roaster is rated for a minimum batch size of 300g, during the drum seasoning I found that I could still get consistent measurements from batches down to 100g (I didn't test any batch smaller than this). While I already had a sample roaster, it used electric heating elements, was very slow to start up, difficult to control, and the chaff collector was too small so a gas fired roaster with a larger chaff cyclone makes a nice upgrade.

Second, it could be used as a research platform. The hardware looks like it's relatively easy to hack around with and while I probably won't have time to do anything along these lines this year, it seems likely that I'll be able to try out some ideas I've had regarding control systems that would have been more difficult to attempt on my other roasters.

The third use case is what this roaster is really meant for, but the least certain. That is, developing roast profiles and having the ability to transfer those profiles to the production roaster. This would let me develop new roast profiles using less than half as much coffee as I've previously used for this purpose but it quickly became apparent that this would not be straightforward because measurements on this roaster were not equivalent to measurements on my production roaster.

I recently wrote about the use of linear spline interpolation to produce a mapping function that takes a measurement from the IR-1 and produces an IR-12 equivalent measurement and modification of Typica to produce this data series. While early results investigating this approach showed promise, I wouldn't know if this was a practical approach until I tried it.

Yesterday I roasted a batch of coffee (a new lot of Mexican Chiapas) on the IR-1, pulling samples during the roast as seen in this video. I started pulling samples much lighter than I normally would because I was also interested in possibly finding a standard sample roast.

This morning I cupped the samples, found one that I wanted to replicate as a production roast, and also found one that seemed like it might be a good choice as a standard sample roast. Going back to the data, the one I picked for possible use on pre-purchase samples was very close to what I was doing on the old sample roaster. As for what I wanted to try for a production roast...

Profile translation was enabled for this batch on the IR-12, but I only missed the translation point by 1 second, so that isn't readily apparent from the graph. I stopped the roast based strictly on temperature measurements (attempting to match the magenta line to the cyan line with chemical change starting at 300°F). Bean color matched with what I had left over from the cupping, and after 5 hours of rest I brewed a pot and found that it matched with my expectation. I'll brew another pot tomorrow before this is allowed out for sale, but expect that I'll still find it delicious. I'm going to tentatively say that this is a viable approach. I have some other new lots that I need to figure out in the near future and I'll take the same approach with those. That will get me a more diverse assortment of roasting styles for testing this.

While this seems to work on two roasters that were designed to be similar, it is an open question if this approach can be used with roasters with more significant design differences. I think it's likely to still be a useful approach, but perhaps not in quite as direct a manner as I've observed here.

[appliedcoffeetechnology]

When discussing roast profile development with novice coffee roasters there were a few areas where my explanations were frequently not understood. I thought it might be easier to understand if I just showed what I was doing in a video and I've finally gotten around to finishing that. It was a bit of a challenge as my laptop is working on dying (that didn't take long).

You might need a bigger view to see all the details. You can get all of the important stuff without that, can see most of the details in 480P, and should be able to easily see all of the details in the 1080P full screen version. That said, the numbers don't matter so much. The topic isn't the profile itself, but the method of finding it. Roast profiles you develop yourself might vary from the one seen in this video considerably due to different coffees, different preferences in what you want to get out of that coffee, and differences in the design of the coffee roaster or in the thermocouple positioning.

[appliedcoffeetechnology]

Lately I've been working on putting together a video on roast profile development. I had hoped to have this finished by now but among other things demanding my attention, my lack of experience working with video editing software, and that my first attempt at narrating the video really sucked, it'll be a little while longer before that's available. Aside from the audio (which was fine technically, but terrible as a performance), everything else is coming together nicely. I'm reasonably confident that will be out before the end of the month, but I'm trying for getting that done before the end of the week. We'll see how that goes. As a preview, I suppose I can disclose a rough outline of the video.

Brief introduction

What is a roast profile?

Examples of roast profiles (ways to make them)

Why use roast profiles?

Why develop roast profiles?

Roasting the coffee

Pulling samples

Cooling samples

Cupping

Adjusting a roast profile

Replicating a target roast profile

Defining QA parameters

There's a lot there and I'm focusing more on what's standard procedure for me rather than going into depth on all of those points in an attempt to keep the video down to an acceptable duration. I'm using Typica to illustrate the procedure, but I'm trying to keep it as a background element as the techniques should all be usable with other software (unless that software is grossly deficient) or manual logging.

Tomorrow morning I'll be on a conference call related to developing a class covering similar material that I'll be helping to present at the SCAA conference in April (RP120 profile roasting if you haven't signed up yet). There will be another meeting about that in February while I and some other involved parties will be in Portland for the instructor development program. I'm excited to be involved in this and hope that people taking the class will be able to take some new understanding back home to improve their roasting operations.