This page created with Open Office Writer, Open Office Draw and KompoZer Best viewed with Mozilla Firefox My Reefer Door - 10K Roaster RETIREMENT: 11 years old forever..............
10K Roaster Page
updated: |
|
This work still in progress... My home-built, big-batch roaster to possibly start a fairly-traded coffee-roasting co-op. The co-op concept will provide coffee that is economically sustainable and fair — for both growers and consumers. The working title is: Fair-Squared Coffee Roasting Co-op I digress. Back to the roaster. Project Goals: ● Try to push this design up to a 37.5# roaster (17Kg, 1/4 bag). But the goal is 22# (10Kg). ● Very low cost. ● Extremely easy to build, with easily purchased materials and little or no machining. ● Usable in a suburban home setting in terms of power, smoke, noise. ● Profile controlled in time/temp. Easy to use. ● Easy maintenance, which in coffee roasters means easy to disassemble and clean. (parts fit in a dish washer?) ● Very reliable. Since I believe everyone who starts using a coffee roaster becomes very dependent on its products, financially and/or chemically. ● Roast small batches (1lb, 450g) with little or no modification.
a vertical wind tunnel - sky diving practice chamber. I'm including some engineering and math details in case someone else might be helped by starting from a calculated first guess. Personally, I hate building something that doesn't work, which then eats up more time and money to make it work. Worse yet, I hate building something that works, only to have a good buddy tell me two years later, "So, why did you build it that way? If you'd done ____ it would have roasted twice as many beans!" Having said that, don't assume that what I calculate is anything like a good guess. I am not an expert in coffee roasting, I'm an electronics guy. But maybe there is something useful here in the references or the reasoning (dreaming). I would very much appreciate any feedback on corrections or explanations for what's presented here. I enjoy the learning process. All the math here is non-trig and non-calculus. To understand and use this write-up, all the reader needs is to dust off the old high-school algebra. However, under no circumstances do I want to ruin all the fun of just building something and trying it. To hell with all the math! Got a fire extinguisher? I sure do. Thinking through the problem... The fundamental benefit of high-speed air heating is the dramatic increase of heat conduction into the beans. The results are faster roasts and/or larger batches and/or lower roast temperatures. In a typical drum roaster, 7" diameter revolving at 60 RPM, the relative bean-air speed is only 1.8 ft/s, at least 20 times slower than useful air spout speeds. Starting conditions: a tall cylindrical roast chamber (RC) of diameter DRC, open at the top and bounded on the bottom by a conical funnel of angle θF. The funnel stem offers a inlet port (aperture) of diameter DIN. The RC is fixed vertically and filled with beans to a depth of HB. The funnel is necessary to ensure that all beans move toward the spout area. To achieve the most even heating, I'm assuming the best RC configuration is symmetrical about the air flow (centered vertical axis). I know some air roasters use an asymmetrical off-vertical chamber to lower the required start-up pressure with excellent roast results. But for now I'll stick to a chamber that's symmetrical about the vertical axis. It's easier to draw the pictures. 
Another assumption I'll make is that heating will decrease away from the center spout, at the outer circumference of the RC, farther away from the central air flow, nearer or in contact with the walls. This loss in temp should be most pronounced near the top rim of the funnel creating a cold zone. Above the bean pile, differences across the diameter of the RC would be minimal depending only on heat losses through the cylinder wall. (I expect to wrap the heater/funnel/RC with thermal insulation to minimize cold spots and heat loss.) Having a cold zone might be a good thing, to allow some dwell time for the heat to penetrate deeply into the bean. But it may also just be extending roast time unnecessarily. A different RC design that can reduce or eliminate the cold zone might be worth a try. Possibly a more steeply angled funnel would minimize the cold zone. Or more radical, would it be possible to suspending the entire batch in the air flow. Either way the penalty for this is a higher bean pile, requiring higher static pressure from the blower. A very deep bean pile would create a bean plug where starting pressure would be forced to lift beans across the entire diameter of the RC. This could easily result in pushing the plug up the RC. A bean cannon? So there might be a minimum practical ratio of RC diameter to bean pile depth. An unusual approach might be to fluidize the entire bean pile with a large inlet aperture and higher air volume, then terminate the bean cloud with a steep funnel to a large diameter stop chamber that slows the air flow and allows the beans to fall back into the narrower RC. This might offer faster roast profiles since the beans would have almost no cold zone or dwell time away from the hot air. The batch-lifting scheme could be supported by a thermocouple (TC) in a small slanted catch-trough in the stop chamber. A stream of beans would be continually caught and roll out slowly over the sensor. Also the blower would need to be larger, optimized for high static pressure as well as more air flow (CFM, ft³/min). Fluidizing the entire bean pile in a roast area between the two diameters might be easy, or it might require some careful balance between the RC diameters DSC/DLZ and the air speed. An optical sensor may be needed to control the air flow by detecting that the beans have lifted off the hold screen. Diameters: All air spout RCs have two diameters: a small one as an entrance port for the hot air where the air speed is fast enough to lift the beans; and a wide segment where the air speed slows down to allow the beans to fall back. Since the same air volume is passing through all parts of the air path, the air must move faster through the smaller diameters. For a given blower output (ft³/min), it is the diameter (actually the cross-section area) in the air path that sets the speed of the air. If the inlet diameter is too large or the air flow too slow, the beans will not spout. Conversely, if the larger RC diameter is too small, the beans will tend to be blown all the way out through to vent. So the two different diameters are chosen both to spout and then to stop the bean's upward movement. Stop Screen: Placing a screen over the top vent can prevent beans escaping the chamber. But if the vent is a smaller diameter than the upper RC, the air will speed-up entering the vent and may hold beans against the screen blocking the air flow. As a precaution, I'll probably use a large diameter vent or mount the stop screen in the large diameter area of the RC to prevent any high-rising beans from getting carried out by fast air. Air Speed: The lifting power of air is especially affected by it's speed, more so than any of the other factors such as bean weight or air density. The equations show that the lifting effect is based on the air velocity squared (V²). If a bean stays the same weight but gets 2x larger, the lifting effect of moving air also is 2x greater. But if the air velocity gets 2x faster, its lifting power becomes 4x greater. A small change in air flow can have a large effect on lofted beans.
My standard small green bean (Yemeni): 0.13g/bean (~3500 beans/lb) 0.14"T x 0.24"W x 0.32"L half prolate spheroid? (0.36 x 0.610 x 0.81 cm) minimum cross-section area = 0.027 in², 0.17 cm² Cd = 0.1 (Smooth oblate spheroid used as low-side estimate to determine worst case maximum air flow.) Compare to drag coefficient of a bullet is 0.3, rough sphere is 0.4, and a solid hemisphere is 0.42+. See Wikipedia drag coefficient table, and Free online textbook Drag of Blunt Bodies and Streamlined Bodies, and EngineeringToolbox Drag Coefficients. converting air density: lb/ft3 = 62.37 x g/cm3 ( g/cm3 = 0.01603 x lb/ft3) Calculate static air pressure needed to lift beans at start-up... 
I modeled the start-up problem assuming the beans have already been loaded before the blower starts up. Certainly there are benefits to starting the blower first, since it avoids the whole problem of static pressure initially lifting some large part of the bean pile to create the narrow tunnel for fast moving air. But designing the physical roast chamber so it can be loaded in place can add another level of complexity, and if the roaster can be built to do a static lift then it can be loaded either way. So I'll try for it. Considering the beans in a pile, I'm assuming that the air from the funnel aperture first needs to lift a cone-shaped area of the bean pile. The area of the cone has little to do with the angle of the funnel, and more to do with how the beans stack one-on-the-other. Taking samples from my bean stash, I did some slump tests to measure the angles of self-supported slopes in bean piles. The typical slump test used for concrete and other materials is something like a beach-bucket test, where some of the material is scooped up in a tall cone, the cone is quickly flipped over onto the ground and lifted carefully up. The material slumps down and the hight of the resting pile is measured. In this case, however, I'm interested in the angle of the slope, not the hight. I'm sure if I knew any civil engineers they could tell me what this test is really called. But for now, the name "slump test" is all I can come up with. Dumped in a pile, green coffee beans slump to a fairly flat angle at perhaps 25°. This naturally varies with the general shape of the beans in each different batch. However, this low angle slope is always the result of beans in motion, collapsing until they stop themselves. When pressed carefully with a flat surface in non-moving tests, the bean pile could sustain a much steeper slope undoubtedly due to static friction being greater than sliding friction. Statically, the beans could sustain a slope of about 53°. I'm going to assume that the complement of this angle, 37°, is an acceptable estimate for the cone angle of beans that will need to be lifted in the roaster. The actual lift volume is an up-side-down truncated cone. I start the cone at the edge of the air inlet and draw it upward from there. So the lift zone is a cone with a small portion at the point removed that exactly fits the aperture diameter. In estimating the lift weight, I'm going to ignore the small added volume at the point of the cone that passes below the hold screen. So it will raise the estimated weight a little, but that's conservative (lazy) engineering. Sin(37°) = 0.6018 Cos(37°) = 0.7986 Cone volume = (π r³ h)/3 If the bean pile is deep enough, it may reach or exceed the point where the lift zone angle intersects the walls of the RC. If so, the lift volume has to include the maximum cone area plus a cylinder volume of beans above the cone. But I believe that when the bean pile exceeds the volume of the cone, lifting the mass to start the spout gets a whole lot harder. Limits on the batch size: bean pile hight? As the batch size increases, the bean pile gets higher, the beans fill the funnel and begin filling up against the vertical walls in the RC. This presents a more difficult problem than just lifting a cone-shape in a pile where beans have room to slide back off the rising lift-zone. Contained by the vertical walls, the beans will not only remain in place on top of each other, but the force will tend to push outer beans against the wall, creating a large static coefficient of friction resisting movement, a clog. Even against smooth stainless steel, this could add considerable weight to the static pressure requirement on the blower. 
I believe this could be one of the important effects that limits the batch size in vertical-wall air spout roasters. One possible way to avoid this limit is to simply extend the funnel diameter, creating an RC which is at the bottom just one big funnel. The idea is not new, since the original prototype for the Sivetz air spout roaster was simply a funnel held vertically, left open at the top. The very essence of a carnival popcorn popper. But the funnel angle need not be as flat as is normally used, nor even as wide as my estimated lift-zone (37°) in a bean pile. Even a steep-walled funnel (20°-30° from the central axis) would present a non-clogging bean load. A narrow funnel would still allow the beans to shift outward as they lift upward. And most if not all of the bean pile would be lofted. I'm hoping an important benefit of a narrow funnel is that it approaches the high exposure batch-lifting version (shown earlier) in reducing bean dead-time and bean cycle time between contact with the hot air stream. But given the same batch size, it should allow a using a blower with a lower static pressure. In effect, I'm hoping a tall funnel offers some of the benefits of all three air-spout roaster variations: the classic vertical-walled small funnel air spout; the large aperture batch lifter; and the large funnel roaster. So for a descriptive name that would sound great in German, I'll call this hybrid the “Tall Funnel Batch Lift Air Spout Roaster” (TFBLASR?).
Calculate air speed and flow rate needed for a sustained fluidized bean column... After the static air pressure has pushed a tunnel up through the bean pile, the continuous air velocity in the tunnel must be sufficient to loft the beans. The bean-loft air speed is the same no matter how many beans are involved. It only has to do with the aerodynamics of the bean and the density of the air. Based on the equations for air flow, the hardest bean to loft will be a small, dense bean. So the worst case could be something like a Yemen or Ethiopian bean. And green beans will be harder to loft than roasted since beans loose about 20% weight and get almost twice as large, so they may easily be half as dense. This also means that if I use a constant air flow during the roast, by the time I get through 1st crack the beans will be flying much higher in the RC. NOTE: I had some trouble with Sivetz's numbers and equations regarding "Bean Levitation" (Coffee Technology, p.240). So I started doing my own estimates and deriving my own equations. Not sure what will come of this...
The short explanation of the math is that I want a green coffee bean to rise in an upward air stream. This will happen if the force of the upward air (on the bean) is greater than the downward force of gravity (on the bean). So I solve for the air velocity needed to just balance the forces, where they are equal and the bean just floats. Then I'll build the roaster to make the air move a little faster than that. I may get fancy and calculate the necessary upward velocity to keep certain amounts of beans in the air at the same time. see Wikipedia drag equation and Wikipedia terminal velocity
Solving the standard drag equation for V : Vt = ( 2mg /( ρACd ))^0.5 (terminal velocity equation) m = mass of the bean........................................0.13 g g = 981 cm/s² (acceleration of gravity)................981 cm/s² ρ = fluid (air) density at roasting temp.................0.64 mg/cm³ (see below) A = cross-section area of the bean......................0.17 cm² Cd = drag coefficient (estimated)........................0.1, 0.42 (see Cd above)
Green bean avg float pressure = 0.13g/0.17 cm² = 0.76 g/cm² air density (ρ) at 68°F (20°C, 293K) = 1.225 g/l, 1.225 mg/cm³ (standard at sea level) air density (ρ) at 450°F (232°C, 505K) = 1.225*293/505 = 0.711 mg/cm³ (0.00071 g/cm³) Solving for Vt : (2 x 0.13 x 981/(0.00071 x 0.17 x 0.1))^0.5 = 4597 cm/s, or 151 ft/s) Using Cd = 0.42, then the required air velocity becomes 2243 cm/s or 74 ft/s. The above numbers are worst case estimates for maximum required air speed through the aperture for spouting beans. Of the two, I suspect the lower number is more reasonable based on the estimation of Cd . But I'm suspicious of the values since even the lower one is higher than the speed offered by Sivetz. His text didn't include the details of the calculation, though he was using a larger, more normal bean size and a lower air temp. As an aside, air temp is important because as it heats up it becomes less dense and therefor exerts less force on objects. The effect is that hot air at 450°F needs a speed of 74 ft/s to lift beans, but room-temp air at 68°F would lift the beans at only 56 ft/s. Using a one-speed fan, the beans will spout differently as the roaster heats up. Keep this in mind for later when I talk about using a bean spout hight sensor to control the air flow. Varying air temp is one of several reasons why a strategically placed optical sensor might be a real useful gadget on an air spout roaster. Aperture Size Estimating the size, power and RPM speed of the blower will depend not on air speed, but on air volume in ft³/min. This important flow rate is simply the product of the required air speed and the aperture cross-sectional area. So what size is the aperture? It's tempting to make the aperture small, so a smaller fan will be sufficient to create the 74+ ft/s air stream. But the size of the aperture determines how fast the entire bean batch will cycle through the hot air spout. The aperture is the main source of heat. Most of the bean pile sits in the bottom of the RC, around the air flow, slowly feeding down, waiting for another hot ride. This isn't to say that the beans are cold. But everything else in the RC is going to be cooler than the spout of air. A small aperture means the beans will wait a longer time away from the heat source, heating will be less efficient and roast times longer.
work sheet 0.5" aperture = 0.567 in² = 0.00136 ft² @ 100 ft/s = 0.136 ft³/s = 8.18 ft³/min
1.0" = 0.75 in², = 0.00545 ft² @ 100 ft/s, = 0.545 ft³/s = 32.7 ft³/min
4.0" = 12.56 in² = 0.0873 ft² @ 100 ft/s = 8.73 ft³/s = 524 ft³/min
I need to run some air-flow tests!
Check out a very successful air spout roaster with a similar roast chamber, two in fact! The Scott Brothers have been working on this for many years. This one was built in 2005 (I believe). Scott Brothers Coffee, follow: home>Why Buy From Us?>Art Meets Science. More pictures at bottom of page: home>About Us>Photo Gallery. Photo enlargements and explanatory text will unfortunately only show up using Microsoft Internet Explorer. I just found out about Scott Brothers in Dec 2008. Thanks to Edward (Homeroasters.org) for the link.
|
