With the arrival of the pitot tube and manometer, we were ready to begin taking measurement for our airflow tests. We first mounted the manometer securely to the prototype. It uses a column of fluid as the indicator, so being completely level is important. Once, balanced, the red indicator fluid was added and we attached the pressure sensing lines.

The pitot tube was put into the stream of air flowing out of the duct. In actual operation, the duct would be feeding hot air into the roasting bowl, which is the primary method of transferring heat (convection).

We then moved the pitot tube to various points in the duct to plot the air speed. It is not uniform across the outlet of the duct, which is important to understand. We also throttled the amount of air to see how the beans reacted to increasing air flow.

The pressure difference from the pitot tube translates into movement of the red fluid in the manometer. The tube is scaled to convert the differential pressure into an equivalent air speed in feet per second.

We plotted the airspeed at various points in the duct out let. It averaged 2,600 feet per minute, which is equal to 30 mph. The airflow is then calculated to be 950 CFM.

We have now completed our initial work in proving the concept using the prototype. It is now time to move into building!

The initial testing of the roaster prototype with airflow added went very well. However, all of the observations were strictly visual and subjective. It is important to take accurate measurements to be able to make design decisions. Therefore, we ordered the necessary parts from Dwyer Instruments to do the testing. The primary measurement we need is airspeed or air velocity. This is commonly done using a pitot tube.

Most people have seen a pitot tube at the airport, as it is the way that even the most advanced airliners measure airspeed. On private planes it may be on the wing, airliners usually have several and are near the cockpit.

The pitot tube is pointed into the airflow. It has a hole in the front of the tube and a series of holes on the side. The front hole is used to measure the ram air pressure or total air pressure of the air flowing by. The side holes measure the static air pressure, which is the pressure of the air if it were not moving.

The pitot tube has two pressure taps connected to the holes measuring the two pressures. These are connected via tubing to a manometer.

The manometer measures the difference in pressure and is calibrated to then give the corresponding airspeed. The greater the differential, the greater the airspeed. Obviously with zero airflow the pressure difference is also zero.

We will be using this test setup to measure the airspeed at various point and with various settings in the bowl. This will enable us to quantify both airspeed and airflow. The higher the airflow, the faster heat can be transferred into the coffee beans, so knowing the maximum that maintains proper bean dynamics is very important. Testing shall begin shortly.

With the holiday season, our pace of development slowed a bit. Even so, we were able to perform our initial airflow testing. We cobbled together a very effective test jig using a Dayton blower, ductwork fabricated by local metalworking company: Typhan, and a hydraulic table.

We assembled the test jig in a garage in Agritopia® and began our initial testing. First of all, we lined up the output duct over the center of the bowl. The duct size is 8″ diameter, which is a very common size in the HVAC industry.

We then started the Dayton blower. Driven by a 3/4 HP electric motor, it can deliver up to 1600 CFM. By means of diverter valves, we are able to throttle the output of the blower.

We then ran the prototype with a full batch of roasted coffee at our standard 90RPM while observing its behavior. We adjusted the airflow from zero to full output. It was clear that any higher output would be too much.

Our next step will be to add airflow and pressure sensors in the ductwork to be able to quantify the airspeed and total airflow.

Happy New Years!

The next major test for the prototype is the addition of airflow. We are using hot, moving air to heat the coffee beans via convective heat transfer. This coupled with a small amount of heat transfer from conduction and radiation is the best way to roast coffee for full bean development and flavor complexity.

The device that generates the airflow is a blower. Our test blower has arrives in pieces, which is handy for showing each component by itself. The heart of the blower is the impeller, which is essentially a centrifugal fan. As it spins, air is forced out radially from the blades and air is sucked into the center of the impeller. This impeller has straight blades and is made of cast aluminum. Impellers also are made with blades that curve in the direction of rotation and also in the opposite direction. We will determine which blade shape is best for our use.

The impeller is powered by an industrial motor. For testing purposes we are using a 3/4 horsepower motor that uses standard household 110V power. In the final design, we will likely use a higher voltage motor.

The impeller is contained within a housing to channel the air. In this case, the housing is made of sheet metal, but others we will consider are made of cast metal. The air enters though the center hole and then is collected and sent out the rectangular duct on the side.

We are designing ductwork to match up with the rectangular duct on the blower. It will have the capability of throttling the airflow and we will attach various probes to measure pressure and airflow volume.

We have successfully tested the bean dynamics for coffee and cacao without any airflow added. The basic concept of the bowl with return vanes works. What has yet to be sorted out is how to evacuate the bowl quickly and completely and how airflow will affect the bean dynamics.

The roasted beans need to be evacuated from the roasting bowl in just a few seconds after reaching the optimal degree of roast to prevent over-roasting. We tested the output of each return vane to be able to calculate how much arc of the bowl edge would need to be opened up to allow the beans to be flung out of the bowl using centrifugal force. The test showed that each vane pitches 0.5 lbs. of coffee per second and there are 14 vanes pitching a total of 7 lbs. per second. With a maximum batch of 35 pounds of green coffee, which is 30 pounds of roasted coffee, our original idea of opening three vanes-worth of arc has an evacuation time of 20 seconds — too long. If we used the entire circumference of the bowl, the evacuation time would be 5 seconds, which is excellent.

This lead to the idea of lifting the entire ring of vanes allowing the beans to be channeled into annular chamber around the bowl and redirected downward into the cooling bowl below. I sketched a possible design for this idea.

We also wanted  to test the how bean dynamics might be improved by moving the return vanes to the the joint between the spinning bowl and the stationary mounting ring for the vanes. This would eliminate some drag before the beans hit the vanes.

We had the vane strip reworked  and then tested it with roasted coffee. There was some improvement in flow. We also noted that the smoothness of all transitions is critical in keeping bean flow uniform.

The next tests we plan on undertaking is painting a few of the beans and running them in unpainted beans while recording their movements with a digital movie camera. We want be sure that the mixing is very thorough and that there are no sticking points where overheating could occur.

Meanwhile, we have ordered the radial blade blower to begin building our airflow test jig. This particular blower has a 3/4 hp motor and can produce a maximum flow of 1600 cfm.

The jig will be moveable and adjustable for airflow (volume and velocity) and delivery height above the bowl. This should allow us to optimize our design with the addition of airflow.

Aloof Coffee

Aloof Coffee is the logical fourth wave in coffee. For too long, we have continued down the path of roasting coffee, even if in third wave it is barely roasted. Roasting coffee, no matter how light, dramatically modifies green coffee in ways that destroy terroir (distinct characteristics of the growing area). At Aloof, we never roast coffee – we let green coffee speak for itself with its grassy, herbaceous notes. This convergence of the burgeoning Raw Food Movement with specialty coffee is truly the fourth wave.

The Aloof Experience

At the Aloof Coffee Bar, we value customer service. The most important thing we can do for a customer is to educate them. Therefore, we have trained our baristas to read facial language and upon any hint of puzzlement or mirth, they tell a condensed 5 minute story of Aloof’s principles of coffee:

1. Roasting is a crime against the farmer. Who are we to cook the goodness out of their product, bending it into a different product?
2. Roasting destroys terroir. How can we truly taste the original product with all of its indigenous notes if we have roasted them out, replacing them with foreign notes?
3. Roasting wastes energy. The roasting process burns fossil fuels, particularly the afterburner used to clean up the roaster exhaust.
4. Roasting takes away nutrients. The vitamins and enzymes found in green coffee are destroyed by heating.

While this may embarrass some customers, we feel that shame and guilt are powerful motivators for change. Once educated, they can appreciate or, at least partake of our unique coffee beverages. Furthermore, we have designed the store to appeal to the psycho-demographic group known as “hipsters” and our approach is hipster approved.

Aloof Beverages

Green coffee is very different from roasted coffee. The notes created by harmful roasting include unnatural notes like berries, stone fruits, citrus, and chocolate. Green coffee has lovely notes of grass, alfalfa, and spirulina. We buy micro-lots from farmers, so that one can taste the nuances, such as milk thistle, oregano, and lamb’s ear.

Since heating destroys enzymes, we do not use heat in excess of 140F. Instead of V60 brewing cones and hot water, we use a Vita-Mix to blend green coffee with fresh spring water and the strain it through the V60. French presses are also available. We serve the coffee in wine glasses to showcase the emerald hue and to help people realize that coffee is like wine in its complexity. (Note: We do not offer Kopi Luwak, due to our sub-pasteurization temperatures and the threat of e. coli)

We do not use an espresso machine for obvious reasons. However, we do make drinks in the espresso bar style. We use two kinds of milk. Goat milk is our favorite as its barnyard notes meld well with the pasture-like notes of green coffee (we also like how the discovery of coffee was with Ethiopian goats nibbling on raw coffee cherries). We also offer soy milk as it is sort of a sister product to how we “brew” our coffee. The milk can be frothed using a mechanical frother, rather using harmful steam. We do not offer sugar, as it does not enhance the flavor of grassy drinks. Of course we do not offer decaf and have a separate 5 minute story we tell customers who ask for it.

In the spirit of sharing, we have even prepared a test beverage for Starbucks, which is our take on a green coffee Frappuccino. It is our hope that they will adopt our approach one day.

[Disclaimer: This is NOT a real concept. I am fully devoted to roasting! Just a thought provoking bit of humor.]

Once coffee has been roasted to the correct degree, the coffee must then be cooled quickly. In virtually all roasters, this means emptying the roasted beans from the roasting vessel (usually a drum) into a cooling section. The beans are then cooled rapidly as the roasting vessel is refilled with green coffee to start the next batch.

In the bowl roaster, I have decided to use stacked bowls: one for roasting and one for cooling. In the roasting bowl, hot air is circulated through the beans, in the cooling bowl cool air is circulated through the hot beans. Ideally, the bowls are not spaced very far apart because there are two human factors to be considered. First, I would like for the roasting bowl to be viewed without the need for steps, a riser, or a ladder. Second the cooling bowl needs to be able to empty into a container not more than 2 feet high that sits on the floor.

There are three ways I have considered for moving the roasted beans from the roasting bowl to the cooling bowl below. The first way is to pivot the bowl and dump the beans into the bowl below. I’ve eliminated this method because it increases the required bowl spacing and makes the drive mechanism more complicated. The second way is to use gravity by lifting the diverter cone at the end of the roast cycle and letting the beans drop into the bowl below.

The third way is to use the centrifugal force of the spinning bowl (already the means for mixing and bean movement) to spin the beans out of the bowl through a duct and down to the cooling bowl. This would involve retracting a section of return vanes to create the opening for the duct.

The duct would have vanes that help direct the flow of the beans in a relatively gentle manner.

Each method has its merits and problems. Dropping coffee through a hole in the bottom of the bowl is simple. The problems are 1) the bowl must be slowed down which may require braking, 2) the cone actuator mechanism goes through the area where the bowl support bearings and drive gear are, which complicates things, 3) the bowl shape for best airflow may not be the best shape to allow quick flow of the beans down the hole, and 4) a scraper blade may be required for speedy flow and that will require an actuator, which is a complication. The centrifugal duct method allows the emptying to happen while the bowl is spinning (no requirement to slow down). The problems are 1) it is a more “violent” process which could be hard on cacao, 2) it requires a mechanism to retract a  section of the return vane ring, which is a complication, and 3) it may not empty as fast as may be desirable.

We plan on trying to modify the the prototype to be able to test the centrifugal duct method. Due to the way the prototype is built, it is impossible to test the drop method, but we can observe bean dynamics that will help us project how it might work. It is possible that we will choose one method to empty beans into the cooling bowl and the other method to remove the beans from the cooling bowl.

One of the major goals of the roaster project has been to design a roaster with enough flexibility to be able to optimally roast a variety of products. The most important is coffee, of course, and if designing a multi-roaster in any way compromises the ability to roast coffee, I will abandon the idea. However, it seems as though by varying certain parameters, one could also roast cacao, malted barley (for beer or distilled spirits), and nuts.

The first alternate product that I wanted to test for dynamic behavior in the prototype is cacao. Cacao is roasted at a lower temperature and for a longer time than coffee. It is also more delicate in terms of physical structure.

We started by cleaning all of the coffee out of the roaster bowl.

Next, we added 35lbs. of mixed, unroasted cacao “beans” to the bowl. This is the same batch size we consider to the be the nominal maximum.

We ran the bowl and found that the cacao behaved very similarly to the coffee with an optimum rotational speed of approximately 90RPM. The flow was very even and it appeared that the cacao was not being fractured in the process. This was something cacao experts who follow this blog had been concerned about.

We stopped the bowl and inspected the cacao for chipping and fracturing, which we found no evidence of. While this was just a single series of tests, we were encouraged that the roaster has the potential to roast cacao.

Special thanks to Waleed Rahhal, Chief Technologist (and Owner) of Fiber Concepts. Waleed is an applied mathematician who analyzes all sorts of components under stress and thermal changes for high tech companies and the aerospace industry. He is also married to my wife’s cousin! Waleed was kind enough to undertake analysis on the roaster bowl. The following work is entirely his.

I’ve been wanting to know how the roaster bowl will expand in size when heated. This knowledge is extremely important to make sure that the fixed components and the rotating components maintain a proper gap throughout the entire temperature range. If we make the gap too small at room temperature, the bowl could rub or even seize at high temperatures. However, we want the smallest gap possible to prevent coffee beans (or coffee bean fragments) from entering the gap.

Waleed first did a hand calculation to determine how the bowl would expand. He used the geometry of the bowl and the material characteristics (for 455 steel), along with the temperature range to do the calculation.

The results show that the radius of the bowl will grow by 0.048″ in going from 70°F to 500°F and that the height of the bowl will increase by 0.024″. Waleed then used finite element analysis to confirm his results and give a better graphical presentation of expansion.

Waleed also added the effects of 35lbs. of green coffee to the model. The results were identical to the hand calculation radially (0.048″).

The finite element model’s results were identical vertically at 0.024″. Waleed also modeled various kinds of stresses in the bowl as it is heated and spins.

This is a graphical representation of the radial stresses on the bowl spinning at 100RPM at full load and at 500°F. We are still reviewing the various stresses in the bowl, but it appears that none of them represent a serious problem.

The modeling of thermal expansion has given us excellent data. We can now use this information to begin designing the fixed components for the roaster.

We finally recovered from walking pneumonia and were able to get back to active work on the prototype. It had been waiting for us at AMCE Metalworks for about four weeks.

We headed over to ACME with a trailer and had them fork lift it into place.

We then transported it to Agritopia, where we used one of our tractors to fork lift it out.

The prototype was then moved into our test facility. We clamped the water jetted return vane strip into place using long reach C-clamps. They were positioned at the margin between each vane. Due to the variable gap and run out between the bowl and the support ring, we fashioned a heavy paper sleeve/gasket to make sure no beans leaked out once the bowl began to spin. As mentioned in earlier posts, the gap and run out problem is a result of unrelieved stresses in the original bowl that manifested themselves when we separated the bowl from the ring.We bent the return vanes into place so as to direct the beans back to the center of the bowl. At this point, it was time to do an initial test, so we put 5 lbs. of roasted coffee into the bowl.

We slowly increased the speed of the motor and as we did, the band of beans crept up the side of the bowl. At about 90 RPM (bowl speed), the beans started to contact the vanes and flowed back into the center of the bowl. Obviously, we were very pleased that the prototype was working well right at the start.

The next step was to go to a larger batch size, in this case 15 lbs. of roasted coffee. That is approximately 1/2 of design capacity. We also began refining the shape of the return vanes for improved performance.

The vanes are designed with tabs to be fashioned into a compound curve. The revised shape did a better job of directing the return flow of the beans.

This test ran very well. A video of the test is shown here: ProtoRun1 – Computer

Over the next couple of weeks, we will  running various batches of roasted and green coffee. During these tests, we will be using both strobe and HD video to observe bean flow, analyzing to see where we can make improvements. The optimum bowl speed for various loading and densities will also be ascertained. We will also attempt to source some cacao to make similar tests. After that, we will add variable airflow from an overhead jet.

We are highly encouraged and feel confident that the bowl roaster approach is sound.