BeanSieve

Competition AeroPress brewing is already built around control. Published World AeroPress Championship recipes routinely specify coffee mass, water temperature, grinder setting, pour structure, stirring, steep time, press time, and final beverage weight [5]. In that context, an instruction such as “18 g of coffee” can look complete.

But weight is only one description of a dose. Two 18 g doses can be identical on a scale and still be compositionally different. One dose may contain more small beans and fragments. Another may contain more large beans. A third may include a wider mix of sizes or broken material. The scale confirms the mass, but it does not describe the physical composition of the beans entering the grinder.

This matters because competition brewing depends on comparison. If a brewer changes grind size, water temperature, agitation, or bypass, they need some confidence that the coffee dose itself has not changed in an uncontrolled way. Otherwise, a sensory difference may be attributed to the recipe change when part of it may have come from a different mix of beans.

Many careful brewers already inspect beans by hand. They may remove broken pieces, very pale beans, burnt-looking beans, shells, or beans that simply look wrong. This is useful. It can remove obvious problems before grinding and brewing.

But visual sorting has limits. It is subjective, it focuses on visible defects more than distribution, and it is difficult to scale across a full competition sample. Some competition rule sets already acknowledge that competitors may sort through beans and weigh doses before official brew time begins [11]. The missing step is not permission to sort. The missing step is a practical way to make sorting measurable and repeatable.

The easiest way to make a sample more uniform is to discard aggressively. A brewer could remove the smallest beans, the largest beans, broken pieces, and anything visually unusual. This may produce a cleaner-looking working sample, but it also creates a practical problem: the brewer now has less coffee.

In competition preparation, that tradeoff matters. If the available sample is around 250 g, every discarded gram reduces the number of possible dial-in brews. This is why Bean Sieve should not be understood only as a tool for removing beans. The more important use is controlled allocation: sieve the full sample, remove the broken below-range material, weigh the retained fractions, and build brew doses from those fractions deliberately.

The key variable in this paper is not simply “bean quality” or “defect removal.” The more precise variable is roasted-bean size profile before grinding: the distribution of whole roasted beans across defined size fractions before they enter the grinder.

Visual sorting is a useful step. A brewer can remove beans that are obviously broken, unusually pale, visibly burnt, badly chipped, or otherwise suspicious. Quakers, for example, are immature beans that often remain lighter after roasting and may be removed by hand or with optical sorting systems [7][8].

But a sieve is not an optical sorter. It cannot look at a bean and decide whether it is pale, under-browned, scorched, or visually defective. It cannot tell whether a bean is a quaker if that bean is the same size as the surrounding coffee. Visual sorting asks whether a bean looks wrong. Bean Sieve asks what size fraction the bean belongs to and how much of that fraction should enter each brew dose.

Bean Sieve uses eight sizing trays plus a bottom catch tray. The catch tray collects material that passes through the smallest 5.0 mm aperture. In practice, this material is expected to include broken pieces, fragments, and very small material rather than intact representative beans.

Once size fractions are separated and weighed, the brewer has a new dial-in variable. They can preserve the retained profile across every brew, remove selected extremes and recompute the remainder, or intentionally bias the profile for a controlled experiment.

These choices are not sensory guarantees. They are recipe conditions. The important point is that the brewer can now make the choice deliberately instead of comparing recipes through a moving baseline.

Coffee has long been sorted, graded, and described through physical properties, including bean size, defect count, density, and cup quality [6][12]. In green coffee grading, bean size is commonly measured with screens. A screen 16 corresponds to 16/64 inch, or about 6.35 mm, and a screen 18 corresponds to about 7.14 mm [3].

ISO 4150:2011 describes a routine method for size analysis of green or raw coffee by manual and machine sieving, and laboratory sieve suppliers describe the same logic in practical terms: coffee is passed through a series of screens with descending aperture widths, and the mass retained by each sieve is recorded as a set of size fractions [1][2].

Bean Sieve borrows the logic of coffee screen sizing but applies it to a different problem: roasted-bean AeroPress preparation. Its current aperture set is specified in millimeters because the tool is designed around roasted beans and practical 3D-printed apertures, not formal green-coffee grading.

Ground-coffee sieving happens after grinding and is concerned with fines, boulders, and grind distribution. Bean Sieve happens before grinding and controls the composition of the whole beans that enter the grinder. The two interventions can coexist, but they are not interchangeable.

Bean size is not roast level, density, color, or flavor. But bean size can still matter because coffee roasting is a heat-transfer process. Beans of different sizes do not necessarily experience the same internal development under the same roast conditions.

Roasters often care about size uniformity because physical size can affect how heat moves through the batch [3][13]. Larger beans can require more time for heat to penetrate evenly, while smaller beans can heat more quickly and may be more vulnerable to uneven development if treated too aggressively.

Bean Sieve should not be framed as a roast analyzer. It cannot tell whether a bean reached the same internal development as another bean. It cannot measure color, mass loss, moisture loss, or chemical development. It only separates beans by aperture size.

That makes size a proxy variable. A proxy is useful because it gives the brewer something observable and controllable, but it is not the thing itself. If a brewer suspects that smaller fractions are contributing to roast-heavy, bitter, drying, or harsh characteristics, Bean Sieve makes that hypothesis testable against a controlled baseline rather than proving it in advance.

The most useful way to think about Bean Sieve is not as a rule system, but as a hypothesis generator. Roast development and roast level can influence cup perception [9][10]. Bean Sieve connects that cautiously to dose composition by making one physical variable testable.

Bean Sieve is an open-source, 3D-printable sieve system for roasted whole coffee beans. It is designed for competition-style AeroPress preparation, especially when the brewer receives or prepares a limited sample of around 250 g and needs to use that sample across multiple dial-in brews.

The tool consists of eight stackable sizing trays, one bottom catch tray, and a live companion calculator for converting tray weights into repeatable brew-dose compositions. The sieve separates the sample physically. The calculator makes that separation usable for dose assembly.

The brewer pours the roasted whole-bean sample into the top of the stacked sieve and shakes the stack so beans move through the trays until they are retained by the aperture that matches their size. Larger beans remain higher in the stack. Smaller beans pass farther down. Material smaller than the smallest aperture falls into the bottom catch tray.

Before sieving, a sample might only look vaguely mixed. After sieving, it can be weighed. That weighing step is what turns the tool from a simple separator into a brewing workflow.

Bean Sieve controls one thing directly: the size composition of the roasted whole beans before grinding. That includes how much coffee sits in each retained size fraction, whether bottom catch-tray material is removed, and whether each brew dose receives the same retained profile.

It does not directly measure roast level, color, density, moisture, internal development, defect status, grind-size distribution, sweetness, acidity, bitterness, or extraction yield. Its value is upstream: it makes the pre-grind bean profile repeatable enough that later recipe comparisons are easier to interpret.

The simplest use of Bean Sieve is to remove the most extreme fractions. The brewer sieves the roasted sample, removes the bottom catch-tray material, and then optionally removes the smallest and largest retained fractions before brewing from the remaining middle profile.

This is quick and intuitive, but it is also blunt. Removing extremes may make the sample more homogeneous, but it also reduces the amount of usable coffee and may remove potentially useful material without testing.

The main Bean Sieve workflow is more conservative. Instead of removing many fractions, the brewer removes only the bottom catch-tray material by default. The retained fractions are then recomposed into repeated brew doses using the same proportional size profile.

This workflow is called size-profile matched dosing. Each brew dose contains the same proportional mix of retained roasted-bean size fractions after bottom catch-tray material has been removed. It gives the brewer a controlled baseline without throwing away large amounts of coffee.

The exact distribution will vary by coffee. Different varieties, origins, screen sizes, processing styles, roast levels, and handling conditions can all produce different tray weights. The example below is illustrative.

In this example, every 18 g brew dose would contain the same retained size profile. If the brewer changes grind size, water temperature, agitation, bypass, or press behavior, they can be more confident that the whole-bean size composition did not drift at the same time.

Once the brewer has a matched baseline, Bean Sieve can also be used experimentally. The brewer starts from the size-profile matched dose, then changes one part of the profile deliberately and tests the result against the baseline while keeping the rest of the recipe as stable as possible.

Bean Sieve creates fractions. The calculator turns those fractions into a repeatable dosing plan. Without it, a brewer could still weigh each tray and calculate proportions manually, but in practice that is slow, error-prone, and inconvenient during competition preparation.

The calculator accepts tray weights for all eight retained fractions plus the bottom catch tray, a target brew dose such as 18 g, and the selected workflow mode. Its first default rule is to remove the bottom catch-tray material before calculating the usable retained profile.

The main output is a per-dose tray plan: usable coffee weight, percentage of usable coffee in each retained tray, grams per tray, maximum complete doses, remaining coffee, excluded fractions, and any adjusted profile after exclusions or biasing.

The calculator should mirror the three Bean Sieve workflows. Matched profile mode preserves the retained size distribution. Removed-extremes mode excludes selected retained trays and renormalizes the remainder. Intentional bias mode lets the brewer change the profile deliberately without pretending to recommend the sensory direction.

A calculator can accidentally make a workflow look more exact than it really is. If it outputs 0.12 g from one tray, that may be mathematically correct within the entered data, but manual bean handling, static, transfer loss, and scale response limit practical execution.

The purpose of the calculator is not laboratory certainty. It is to reduce uncontrolled variation enough to improve practical comparison.

Existing coffee practice supports the first part of the Bean Sieve argument: bean size is a real and measurable coffee variable. Coffee screen sizing is already used in grading and quality control [1][2][3]. Existing roasting sources support the narrower idea that bean size may influence roast behavior [3][13][14]. Existing sensory research supports the separate point that roast development and roast level can influence cup perception [9][10].

But this does not prove that Bean Sieve always improves coffee, that one specific tray is universally better or worse, or that smaller fractions always cause bitterness while larger fractions always cause underdevelopment. The tool supports experiments. It does not provide conclusions before those experiments are run.

Before making sensory claims, the first test should be mechanical repeatability: if the same coffee is sieved repeatedly under the same conditions, does Bean Sieve produce similar tray weights? The second test is dose assembly: can a brewer repeatedly prepare 18 g doses that match the calculator’s target profile closely enough for competition use?

Waste should also be measured explicitly. The strongest practical question is not “does it always taste better?” but “how many controlled brews does each workflow preserve or remove from a limited sample?”

The most interesting question is also the hardest: does controlling or changing roasted-bean size profile affect the brewed cup? That should be tested by comparing a size-profile matched baseline against one deliberate profile modification while holding the rest of the recipe stable.

Blind A/B and triangle tests are useful here because a profile can be different without being better. The most useful community report is not “this tasted better,” but “this coffee had this size distribution, we prepared this matched baseline, we changed this one part of the profile, and under this recipe this is what we observed.”

Professional coffee screens and commercial bean sieves already exist. Bean Sieve is not novel because it uses apertures to separate coffee. Its contribution is the combination of roasted-bean use, AeroPress-specific competition preparation, a 250 g sample workflow, default broken-material removal, size-profile matched dosing, a live companion calculator, and open-source 3D-printable distribution [4].

Ground-coffee sifters address a different variable: particle-size distribution after grinding. Bean Sieve controls the whole-bean profile before grinding. Visual hand sorting is complementary because it can remove visible defects that a size sieve cannot identify. Optical sorting can detect color and visual outliers, but it does not solve the specific problem of proportional size-profile matched AeroPress dosing [7][8].

The strongest use case is competition preparation with a limited roasted sample. The main workflow removes only the bottom catch-tray material by default, then uses the retained size fractions to prepare size-profile matched doses. This allows the brewer to keep most of the sample while making each dose more compositionally consistent.

Bean Sieve is also useful for preparing repeated brew doses in advance, reducing broken-bean material, testing size-profile hypotheses, teaching dose composition as a variable, and generating shareable community data for later refinement.

Bean Sieve does not guarantee better coffee. It does not measure roast level, density, color, internal development, defect status, grind-size distribution, extraction, or sensory quality. It does not replace visual sorting, sensory dial-in, or good recipe design.

• The workflow takes time compared with scooping directly from the bag.
• Very small target weights can be difficult to dose reliably by hand.
• Shaking method, tray loading, static, and user handling can affect fraction repeatability.
• 3D-printed apertures have tolerances and should not be treated as calibrated laboratory hardware.
• The tool is most useful when the brewer cares about comparison, not when making one casual cup.