Modeling the Bowl – Part 1

2010 April 29

Special Note: Our photojournalist for the project, Chanelle Richardson, will be taking the photos from now until the unveiling of the roaster. You may see an occasional photo taken by me, but it will be obvious, I am sure.

The Meeting

Part of the design process is working with people who have specialized knowledge and skill that add to the design. It is impractical for me to do everything myself and learn everything that could be helpful during the design phase. However, it is critical is for me to know which things I should do myself and want to do myself versus things that ought to be subcontracted. An important discipline in modern design is the use of computers to model complex systems prior to prototyping. Modeling can speed the design development process, save cost in prototyping, and help one visualize how different variables affect the design. Jeff Zhang is a PhD student heavily involved with computer modeling at ASU. We wanted to meet with him to discuss the unique problems associated with the bowl roaster.

Lewis Berman asks for clarification on my description of coffee beans dynamics in the roasting bowl

Another important part of the design process is constructive criticism and encouragement. My collaborator, Lewis Berman, plays this role. He has a wealth of knowledge and experience in many areas of science, engineering, and manufacturing.  We meet frequently to discuss all facets of the roaster. He challenges my assumptions and forces me to explain my ideas. This is exceedingly helpful and enables me to more quickly narrow down a design and avoid going down unfruitful paths.

Jeff, Lewis and I met at Liberty Market to discuss how computer modeling of the roaster bowl, particularly the bean dynamics, could help the design process. I first explained the goals of the Roaster Project and then gave and overview of coffee roasting. We then moved to the particulars of a bowl geometry.

A Discussion of the Bowl

The bowl offers many benefits in the roasting of coffee. It somewhat odd that no shop roasters use this geometry and default to a drum. Perhaps there are practical reasons we shall discover upon further research. We did find a roasting plant scale bowl roaster made by Probat called the Saturn. It is designed to roast up to 550kg of coffee in one batch, which is HUGE. We are designing for a capacity of just 15kg and it may be that the bowl design is only appropriate for large bean masses. This is something that computer modeling will help us determine.

Sketching thoughts on the geometry of the roasting bowl, including a center cone for air deflection

I drew for Jeff and Lewis some of my thoughts on how the bowl might be configured. First of all, it is very desirable for the beans to be rapidly agitated and thoroughly mixed. It is also desirable that the flow of the beans allow for plenty of convective heat transfer by directing them into a stream of rapidly moving hot air. Based upon these desirable characteristics and formulas involving centrifugal force it seems like a bowl of proportions similar to a wok makes sense. Steeper walls probably will not get the bean velocity where we want it, as I would like to see the beans flow up the walls of the bowl and curl back in a flow to the center. This requires a fair amount of bean velocity when it gets to the outer edge of the bowl. At this point, either the lip of the bowl needs to curve sharply inward to launch the beans to the center or a stationary rim of similar curvature could do the same.

I plan on having a central jet of hot air blowing down from above the bowl. This air would interact with the beans as they curl over and also mix with the beans as they flow back up the sides of the bowl. The airflow should also help move the beans toward the edge of the bowl.

At the center of the bowl, I have added a cone. The purpose of the cone is to (1) deflect the beans back to the ideal travel path, (2) to deflect the central air jet spreading the airflow out across the bowl, and (3) covering a hole through which the beans drop when the roasting process is done. I would envision this cone popping up or dropping down at the end of the roast cycle, allowing beans to evacuate quickly from the bowl and down a chute to the cooling chamber (yet to be designed).

How to Model

After we had discussed the roasting process and the particulars of the bowl roasting vessel, Jeff introduced various ways in which we could use computer modeling to aid in design. The easiest way would be to treat the coffee as sand and do some initial flow analysis for different shapes of bowls. This would be very helpful in narrowing down designs that we might want to physically prototype. The goal is to come up with two or three very promising bowl shapest. A more rigorous type of modeling would be to actually model the movement of individual beans with the characteristics of real coffee beans. This is much more complex, but may make sense, depending on the outcome of the “sand” flow model.

Jeff Zhang suggests uses for computer modeling

After we make prototypes (my goal is try to get it right with not more than 3 full sized bowl prototypes), computers can also help us evaluate the designs. Some companies have painted 25% of the beans white and then watch them flow within the roasting vessel. Jeff, suggests that we do an RGB test where we paint 1/3 red, 1/3 green, and 1/3 blue. We would then focus a digital camera on the flowing beans and be able to do image processing to get an even better idea of actual flow.

He also suggests that if we want to go further still, bean shaped sensors could be obtained very inexpensively (25 cents each). We would put lots of them into the mass of real beans and start the roasting bowl spinning. Each sensor’s position would be tracked electronically. This would directly measure the path of each sensor bean within the flow.

We will need to sort out which of these items are necessary and which would just be fun to know, since the more complex ones take additional time and money. Nevertheless, computer modeling of the roaster bowl will be instrumental in creating a great design at reasonable cost and on a reasonable timeline.

One Response
  1. Kevin Briggs permalink
    April 30, 2010

    Joe, thank you for sharing your thoughts and experiences!

    After learning about your roasting chamber proposals I have been thinking about a roasting chamber shaped like a paraboloid or hyperboloid-two-sheets (only one of the sheets being used), the “vertex” at the lowest elevation. For more uniform roasting of the beans I have been thinking of moving the beans by rotating the roasting chamber about the “axis”. I have been thinking about supplying “heat” to the beans by “hot air”; the roaster chamber constructed to allow air flow from “outside” to “inside” (e.g., mesh, holes); air directed through “nozzles” from outside to inside chamber to “stir” the beans from the “bean distribution by centripetal force” of the rotating chamber. I have been thinking of loading through a gravity drop and unloading by tipping the chamber (tipping rotating axis near the center of mass, in elevation, of the chamber and beans).

    To allow a variety of energy sources to roast coffee I have been thinking of heating the air delivered to the roasting chamber.. One form would use a gas flame (e.g., natural-gas, propane) in a combustion chamber; flame and combustion-chamber-air-flow would mix with another air-flow stream to achieve the desired roasting-chamber-air-flow temperature. Another form could use an electric-heater to heat the air. Another form could use a “heat sink” from a solar collector to heat the air.

    Thank you for sharing your thoughts and experiences in designing a coffee roaster. Thank you for providing a forum for others to comment and contribute.

Comments are closed.