A systematic approach to roast profiling
As the official launch of the P3000 draws closer, one question we get asked a lot is whether it will be possible to transfer profiles from the L100 Plus sample roasters over to the P3000.
It’s a fair question: after all, both roasters are packed with sensors, giving you an unprecedented amount of information about what’s happening during the roast. They’re designed along the same principles, and both are built at our HQ in Norway to the same exacting standards.
In the future, perhaps we will be able to use machine learning to guide our customers through the process, making it easier to use your sample roaster for profiling on your production roaster. But as we explained in our last blog post, there’s no simple way to take a profile from any roaster and perfectly replicate it on a different one — at least, not yet.
Armed with the high-precision sensors on a ROEST, however, you can use the data from your sample roasting and your knowledge of roasting on your machines to help you predict what kind of profile will work best for a given coffee. With experience and practice, you can use this to make the process of developing a profile much faster — and to keep improving the flavor of your roasts over time.
Jan Komarek, Head of Coffee at Bailies Coffee Roasters, has developed an intricate system for using the ROEST sample roaster to guide his production roasting. At their facility in Belfast, they use a ROEST for sample roasting and profiling, and have three very different Probat roasters for production: a P5 for small batches, a classic 1960s UG22, and a Neptune — a 120-kg capacity hot air roaster. We spoke to him about how he uses the ROEST to help him perfect his roast profiles on these three very different machines
Jan’s colleagues, Finley and Eilís, roasting on the 1960s UG22 at Bailies Coffee Roasters
Exploring the coffee’s flavor
In our previous post, we talked about the ‘flavor map’ — a very simple way to investigate the potential of a coffee. Jan uses a similar approach when profiling a new coffee, but takes it a few steps further. “One of the things that I believe in is when you’re developing a roast profile, it’s good to try all the different variables,” he says.
The first variable he tests is roast degree. ‘This is pretty easy with the ROEST, we have a standard roast curve and just terminate it at different temperatures,’ he says. In a blind cupping, he can then determine what roast degree brings out the best qualities from the coffee.
The next variable to test is the total roast time. With patient trial and error, Jan has developed a system to get the same roast degree with different roast times on his ROEST. “We use a single BT/IT curve and move all the inlet temperatures up/down 5-10 degrees to hit the target roast time and adjust the end temperature to hit the identical roast degree,” he explains. He has created a spreadsheet to calculate exactly what adjustment he needs to make to the inlet temperatures to reach the target roast time.
Jan Komarek, Head of Coffee at Bailies Coffee Roasters, uses the ROEST L100 Plus for profiling as well as sample roasting
“This way we can put eight roasts on a cupping table, from 6 minutes to 15 minutes, and choose which roast time we prefer,” he says. “And then, we can build the next layer where we explore heat application over time: for example, should we use high heat at the beginning, or start with more of a soak?”
At the end of this process, he takes the roast profile he has developed and goes right back to the beginning of the process, to validate the earlier choices. For example, with the particular style of profile he chose, is the initial degree of roast still optimal, or does it now taste better roasted fractionally lighter
Even with the help of ROEST’s automation, the procedure is time-consuming — but well worth it, Jan says. “The results have surprised us, and we’re getting better coffees as a result. What is exciting about the ROEST is exactly this — we can do this now, and have a procedure to do it!”
Translating a profile
After developing the perfect profile on the ROEST, the next step is to translate that to the production roaster. Unfortunately, there’s no way to do this directly, Jan says. They tried to find a way to offset the inlet temperature on their Neptune to match the profiles on their ROEST, but without success.
Nonetheless, the profile they choose on the ROEST still informs the profile they create on their Neptune. “If a coffee benefits from higher heat at the beginning and less at the finish, that’s enough information, even if it’s not exact numbers,” Jan says. “If you look at the shape of the curve over time when we scale it up, the application of heat at different moments during the roast seems to be proportional, even when the profiles on the ROEST are significantly faster.”
While he hasn’t been able to translate profiles directly from the ROEST sample roaster to his production roaster, Jan has had some success in the opposite direction — bringing a roast curve from the Neptune over to the ROEST, to help develop his existing roast profiles.
Perfecting production profiles
“We have some coffees where I’m sure we can make the coffee taste better, but we can’t play around with the profile on the Neptune — so we try to transfer the existing profile to the lab roaster, and play around a bit. Then we can apply those learnings back onto the big roaster, knowing how the coffee behaves when those changes are applied,” Jan says.
To translate the profiles from the Neptune to the sample roaster, Jan made use of the Bean Temperature (BT) profiles on the ROEST. He starts with the same BT points as on the Neptune, and then shifts the whole curve by a few degrees until he ends up with the same degree of roast on the ROEST. “You pretty much get coffee that is the same roast degree, same roast time, and has a very similar shape of roast profile,” he says.
The result isn’t always identical, but it’s the closest that they’ve managed to come so far. Using these profiles as a starting point, they can start to explore how changes to the production profile would affect the flavor.
This technique helps them avoid falling into a habit of always using the same profile when roasting coffees from the same origin. “Sometimes similar coffees lend themselves to very different roast styles,'' Jan points out. It also helps them ensure that their roast profile is still optimal, even as the coffee changes over time. “Sometimes if a coffee tastes like it is aging, we can go back to the lab roaster and check if it’s the green coffee, or if changes are needed to the profile — it’s an easy way to check.”
While Jan’s systematic approach to profiling has greatly helped his roasting, he warns that it’s not possible to just copy the technique on a different roaster. “It’s specific to our machine, we can’t hand it to someone else,” he says. “They would have to figure out what is equivalent on their roaster.”
The ROEST approach to profiling
In the future, we are confident we can make it easier to transfer profiles from the sample roaster over to the P3000 production roaster, using techniques like machine learning and empirical modeling. But the ROEST is already a versatile tool for developing roast profiles on any machine, and its flexibility makes it easy to try radically different styles of roast profile. For Jan, the ability to roast the same coffee on the same profile ten times back to back, changing just one variable such as end temperature, makes the ROEST crucial to his approach. “I’m not aware of any other lab roaster that is able to do that,” he says
“I like the fact that ROEST has a lot of ways to control it — we’re getting very good consistency and repeatability batch to batch: it’s very precise, very controlled,” he adds. “It’s great to have the level of control and information the ROEST provides — and the fact there are 4–5 different ways to roast coffee gives us new solutions to different problems.” One example of this is the ability to use modified BT profiles to copy their production roasts — something that is only possible thanks to the ROEST’s sensitive, accurate BT sensor.
“We’re very pleased with the ROEST, because it allows us to try things out,” Jan says. “If there’s anything we’ve learned from these exercises, it’s the importance of it — we’re looking back and thinking we should have done this a long time ago.”
We hope you found this article insightful; to learn more about our sample roasters and explore how they can elevate your coffee roasting process, visit our website or get in touch with our dedicated sales team today.
