Coffee-roasting variables associated with volatile organic profiles and sensory evaluation using multivariate analysis

Abstract

A multivariate analysis investigated the operational variables of a professional roaster, the volatile chemical composition of roasted specialty coffee and how they correlate with the sensory evaluation of the final beverage, which directly affects the commercial value of the product. The experimental planning elicited 38 distinct roasting profiles that variegate (i) drum initial temperature (160 and 200 °C); (ii) drum air-flow ventilation opening (25-50-100 and 25-75-100 % – numbers show the percentage of opening for ventilation throughout the roasting); (iii) drum rotation speed (40 and 65 rpm); (iv) fuel gas flow/pressure increase into burners (0.5 and 1.0 mbar min−1); (v) roasting time after bean cracking (1.0 and 2.0 min). After roasting, 108 volatile organic compounds (VOC) were measured by SPME-GC-MS in the aroma of the roasted coffees. Principal component analysis (PCA) correlated VOC profiles, the operational variables of the coffee roaster and sensory evaluation. PCA from either sensory or chemical analysis data indicated “time after bean cracking” as the major influence and 30 seconds are enough to diminish a couple of points in coffee quality within the Specialty Coffee Association classification. The interaction between variables “fuel increase ramp” and “time after bean cracking” also altered VOC profiles and separated distinct roasting groups. PCA from chemical analysis delivered better separation between sample sets than PCA from sensory evaluation concerning either “time after bean cracking” or the interaction “fuel increase ramp” versus “time after bean cracking”. The coupling of sensory, chemical and operational data and their interpretation through multivariate analysis can improve the knowledge of how the operational procedures affect the final quality of coffee.