Special Seminar by Asst. Prof. Theeradech Mookum

The precise control of Particle Size Distribution (PSD) in coffee grinding is essential due to its direct impact on extraction efficiency and the sensory qualities of brewed coffee. The grinding process involves the brittle fracture of coffee beans, resulting in a heterogeneous mixture of particle sizes, broadly categorized into “fines” (very small particles) and “boulders” (larger fragments). This multimodal nature of the PSD complicates characterization and necessitates advanced mathematical models to capture the complexity of the particle populations. Accurate modeling of these distributions is crucial for optimizing grind settings and achieving consistent extraction profiles. Traditional comminution models, such as Rosin-Rammler and log-normal distributions, provide foundational approaches but often prove insufficient for the complex, multimodal PSD of ground coffee. To address this, sophisticated statistical estimation techniques, including Non-Linear Least Squares (NLLS) and Maximum Likelihood Estimation (MLE), are employed for precise parameter fitting. Additionally, algorithmic approaches like Expectation-Maximization (EM) and Levenberg-Marquardt (LM) methods are utilized to solve mixture models representing the combined distributions of fines and boulders. These computational tools enable the deconvolution of overlapping particle populations, offering a more detailed and accurate representation of the grinding output, which is important for process optimization and quality control in coffee production.