The acoustics of coffee roasting

Date

2017-05

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

The cracking sounds emitted while roasting from eight different coffee bean varieties were recorded and quantified to determine the correlations between the acoustic characteristics of each roast and specific bean properties or roasting temperature profile. A framework was developed to compare acoustic characteristics between different roasts with the future goal of creating an automated acoustic roaster or using acoustics as a real-time quality monitor. An acoustic pressure amplitude thresholding algorithm was implemented to detect crack sounds emitted during both first and second crack. This algorithm was used to detect cracks in forty different 0.25-kg roasts of eight different bean types and using two roaster temperature profiles. The acoustic pressure amplitude and frequency of first crack and second were calculated and corroborate previous findings that first crack tended to be louder and have more energy in lower frequencies than second crack. The maximum crack rate in the first and second crack, onset time of first crack, and separation time between first and second crack were quantified for each roast and correlated to four different roast or bean parameters. Choice of temperature profile and bean origin had significant impacts on the cracking rates while no significant difference was observed between wet and dry processed beans nor caffeinated and decaffeinated beans. Ten 0.25-kg roasts were conducted of the same Ethiopian wet processed bean using the same roaster and preset temperature profile to measure the possible variability of acoustic characteristics between similar roasts. Maximum cracking rates during the roasts varied by more than a factor of five. The onset times of first and second crack both had a range of approximately 60 s. Minor temperature changes within the roaster showed little correlation to temporal differences in acoustic characteristics and a medium correlation to the number of detected cracks. In a separate experiment, two 0.25-kg roasts of similar beans were conducted except one batch was dehydrated to 93.5% its initial weight. The dehydrated beans had a higher maximum crack rate in both first and second crack, earlier onset times for first and second crack, and smaller separation between first and second crack, than the standard beans.

Description

LCSH Subject Headings

Citation