TUM group fits five-parameter models to the bimodal grind distributions of real grinders
Mahima Bora and Heiko Briesen at the Technical University of Munich measured laser-diffraction particle size distributions of coffee ground on different grinders at several grind levels and from beans of different roast degrees, then compared ten two-component (five-parameter) distribution models for how well they describe the characteristic fines-plus-coarse bimodal shape. Lognormal/Weibull, Weibull/Betaprime and Gamma/Weibull fitted best across the dataset; fit quality depended strongly on grinder type and grind level and only minimally on roast degree. The point is methodological: reducing a full PSD to five parameters lets future extraction and sensory models use the whole distribution instead of a single median particle size. Caveat: no brewing or sensory measurements; the study characterises the powder, not the cup.
Why it matters: Nothing to change today, but it reinforces that a grinder's 'setting' is a poor description of what water sees; the fines-to-coarse shape differs by grinder even at the same median, which is why two grinders at the 'same' size brew differently.