A discussion on roasting between a coffee roaster and a process engineer
BY CLAY BUSH
"You're not using the roaster correctly! If you want to get the best out of it, you're going to have to change the way you think."
These words were from my brother Jake during a conversation whilst planning a retrofit of our small 400gram electric Diedrich coffee roaster. The roaster doesn't fit our MO anymore, so instead of investing in a new roaster, we decided to modify our Diedrich. This led to a very enlightening conversation that took a few hours and a change of perspective on my behalf to understand fully. I think this conversation may be of interest to professionals out there, so I have written down the bones of this conversation in the hope it provokes some thought from others.
"The problem is that roasters (machines) don't have enough data inputs to make the PLC work correctly. So you're forced to use it in the wrong manner."
What Jake was saying is that our industry is not using our roasting machines as we should. Initially, like all roasters, I was taken back, but eventually, it started to make sense.
But let's back up a bit and talk about why my brother has the authority to make those statements.
Jake would be the first to admit he is not a coffee roaster and freely admits he doesn't know how to roast coffee. His job was as a heavy industry process engineer for 20 years. He has maintained and programmed industrial-sized ovens to create everything from aluminium to chocolate. But why listen to his views on roasting? To understand that, we need to delve deeper into what a process engineer is.
What does a process engineer do?
A process engineer, as the name implies, is an engineer of processes. This process is another way of explaining the task of taking a raw material and building an end product using machinery and PLC's (programable Logic controllers) to get the desired result you want in the most economical and, hopefully, environmentally friendly way.
So what is a PLC?
A PLC is a control system that monitors the state of inputs. For example, time and temperature, BTU, fan speed etc. then, process this data, which triggers outputs based on pre-programmed parameters to achieve the desired result.
Unless you are roasting on an old school roaster, with all manual controls, there is a high probability that your roaster has a PLC. It is located in a cabinet, most likely under your control panel. Below is a typical photo of a PLC.
Whilst there is a 95% chance you have a PLC, there is also a 95% chance you're not using this PLC in your roaster as you should be!
Not too many roasters I know could explain to you what a plc is, so they definitely wouldn't have the specialised skills to program it. Not to mention many PLC's have manufacture's access codes on them to stop you from altering them. However, programing a PLC is the bread and butter of a process engineer. So whilst my brother may not be qualified to roast coffee, he is apt to tell me I don't know how to use the roaster. After all, a roaster is just an oven. And with only gas and fan, not a very complicated one.
So how is it we are using the roaster in the wrong way?
"As long as your PLC is programmed correctly, it has the potential to be smarter than you and roast coffee better than you."
I have yet to see a PLC that can pre-empt and navigate a first crack crash even if it wanted to. Because of this, roasters have taken the view that we cant use automation to roast coffee but instead must control the coffee roaster. We have learned to use the roasters by changing gas and fan settings. In doing so, you are not utilising the PLC in the manner that it is designed. Instead, you tell the PLC what to do, and the PLC then makes changes to the roaster. It's a very simple process roasters use the world over. But it's fundamentally incorrect.
To illustrate, let's use the analogy of driving a car.
We operate a car by using an accelerator and break to get to a destination. At the risk of using a simple analogy, a roaster is the same; instead, we use gas and a fan to get to a destination. our time-temperature curve.
However, imagine if you could tell a car where you want to go under what circumstances and then the car delivered you safe and sound. This would be a driverless car. A PLC used correctly is a driverless car. You tell it where you want to go, under what circumstances, time, maximum speed etc., and it takes you there. Currently, roaster PLC's can't do this because our roasting machines don't have enough inputs to ensure the PLC can drive itself.
As a result, no PLC that I know of can mitigate a first crack crash. Because the "car can't drive itself", we are left to drive the car, and the PLC is not used as it should be. And around and around we go.
The only way to break this cycle is to make the PLC more capable of roasting the coffee automatically. To do this requires a new way of thinking, a reverse way of thinking.
"So what is it you want the roaster to do? I can't program it till you tell me what you want it to do."
This is where I hit a wall as a roaster. When asked to explain what I wanted to achieve, I was supplying vague concepts at best. As roasters, we are always looking forward to an endpoint and making the best guesses to get us there. With the lack of a PLC, we are forced to roast in real-time, going ahead and making adjustments on the fly. I was so used to telling the roaster what to do to get to an endpoint that when faced with a question on how I wanted to improve the roaster, it was hard to answer. It's hard to answer because regardless of roasting philosophies, we generally follow a time/temperature curve to get to a destination using predominantly gas and airflow to achieve it. However, if you want to use a PLC to its full potential, you must know where you want to get to and how you want to get there! Gas and air are not enough parameters for a PLC to achieve its end goal. So, what other parameters does it need? What would you add if you had an open canvas? This is not as easy to answer as you would have first thought.
The reason is that coffee professionals measure coffee in the resultant flavour. And as such, use roasting machines to develop this flavour. This is a big problem because it’s in opposition to the roasting medium, which functions through heat transfer and physics. So to roast like a machine, you must speak its language. And many roasters can't communicate in this way.
To communicate, to the PLC you must
Articulate in a medium that a PLC can measure.
A PLC cant taste acidity or sweetness, so what in heat transfer makes the sweetness? Because that now becomes your medium of communication.Determine the best unit of measurement for that parameter. For example, temperature. How many different forms of heat transfer happen in the drum? Is temperature the only way to measure them?
What is the range the parameter must function in before it becomes detrimental to the end product? For example, Tipping or scorching on a bean can be from too much heat applied, or it can be from the wrong Bean Mass or both. You must split these factors off and determine boundaries for both of them. These boundaries give the PLC a range it can perform in.
At what stage of the roast is this parameter relevant? Not all parameters will be used for the entire roast. You may roast with air for one part of the roast and conduction at a different stage. The PLC must know when to use what. This is determined by the preset parameters.
If this change of mindset is not challenging enough, it may be even more complicated than we first think? There may be the possibility we have developed our current skills through an inadequate system through no fault of our own. Remember, if all we have ever used is time and temperature, gas and fan, how do you know these were correct in the first place. If all you have in your toolbox is a hammer, everything starts to look like a nail. Furthermore, when asked what more tools would you like, the risk is you ask for five different types of hammers.
Jake tells me the problems we face in the coffee industry have been long solved in other industry’s. In far more complicated scenarios than we face in roasting. We just don’t know it. The trick is coming up with a new standard. A middle ground that can convert flavour to the mathematics the PLC needs to function. ie we must measure the bean by maths which can result in flavour. In this way, we can finally make the bridge and also communicate to the PLC to achieve better-tasting coffee.
Roast Density; the missing link.
Here at the Pillar, we often discuss how Roast Density is so essential to brewing. Rarely tho do we discuss roast density in relationship to roasting. However, in roasting, Roast density is the mathematically defined medium between heat transfer and flavour. We can manipulate heat transfer to modify our roast density which in turn changes our end product flavour, and we can use the PLC to do it. Because Roast Density is determined through mathematics, the PLC can communicate in mathematics; therefore, Roast Density will become a cornerstone resultant in the programming of the PLC inputs. This, together with my brothers' understanding of PLC's and "ovens", gave us the starting point on what modification we would make to the roaster.
The goal is to deliver enough lines of data that the PLC can now utilise this data to make decisions and manipulate heat transfer of its own accord. Our starting point is dividing the roast into five stages and using eight different parameters, with an acknowledgement that we may need to add more. We will keep you posted on how this experiment unfolds.
Cheers
Bush and Bush Coffee Systems.