Steam stripping of aroma from roast and ground coffee: A - PowerPoint PPT Presentation

Steam stripping of aroma from roast and ground coffee: A mathematical modelling approach David Beverly Peter Fryer, Serafim Bakalis, Estefania Lopez-Quiroga, Robert Farr and John Melrose Monday 10th September 2018

Why investigate aroma in instant coffee?

Why investigate aroma extraction? Making coffee ‘more aromatic’ or giving ‘new & improved flavour’ requires process modification Aroma is complex… Thiols Diketones Furanones Guaiacols Pyrazines 3

The instant coffee process and aroma extraction

The instant coffee process Grinding Roasting AROMA Spray Extraction drying Freeze drying Evaporatio n 5

The aroma extraction process – steam stripping Plant scale • A previous Government funded collaboration (Innovate UK) with University of Birmingham was project RICE (Reduced Energy Instant Coffee) Aromatized condensate • Project aim was to reduce energy and water use in the freeze drying process by 25% by increasing ’00s kg coffee the solids content of the feed to the freeze-dryer < 1.8 mm grind size 4-70% w/w added water • Using more concentrated aromas is one way to achieve this concentration increase 1.1-2.0 barg inlet Steam 2-40 mins steaming • Process aims are to maximise yield and concentration of aroma whilst tailoring the 6 sensory profile

The nature of roasted coffee 100 1.8 90 1.6 Cumulative distribution Q3 / % 80 1.4 Density distribution q3* 70 1.2 60 1.0 50 0.8 40 0.6 30 0.4 20 0.2 10 0 0.0 1 5 10 50 100 500 1000 particle size / µm Bimodal size distribution of coarse particles (see SEM) and fragments of cell wall (fines) Mesopores (~40 μ m) Cell wall (~10 μ m) Soluble material & Nanoporous aroma starts here 8 cm 7

Coffee modelling in literature 1) Caffeine extraction characterized by effective diffusion coefficient 2) Extraction from solid phase to intra- and inter- granular pores using mass transfer coefficients 3) Measurements of aroma extraction fitted with apparent diffusion coefficient and/or Weibull distribution 1) Spiro, M. and Chong, Y.Y. (1997). The Kinetics and Mechanism of Caffeine Infusion from Coffee: the Temperature Variation of the Hindrance Factor. Journal of the Science of Food and Agriculture, 74, 416-420 8 2) Moroney,K.M., Lee,W.T., O’Brien,S.B.G ., Suijver,F. and Marra,J. (2015). Modelling of coffee extraction during brewing using multiscale methods: An experimentally validated model. Chemical Engineering Science, 137, 216-234 3) Mateus, M.L., Lindinger, C., Gumy, J.C. and Liardon, R. (2007). Release Kinetics of Volatile Organic Compounds from Roasted and Ground Coffee: Online Measurements by PTR-MS and Mathematical Modelling. Journal of Agricultural and Food Chemistry, 55(25), 10117-10128

Model Description

The modelling strategy Column section O ~ (cm) Fine particle Mesopore O ~ ( μ m) δ Column O ~ (m) P out d pore d part Water film Particle section O ~ (mm) Z Coarse particle Oil film P in Water-filled mesopore d col Nanoporous cell wall 10

Particle scale extraction – the mesopore – aroma’s release from the oil 𝒅 𝒆𝒑 𝑠 𝑞𝑝𝑠 𝒅 𝒆𝒑 𝒅 𝒒 𝑳 Τ 𝒑 𝒙 A spherical oil shell with aroma concentration 𝑑 𝑒𝑝 coats the mesopores Partition into water described by octanol-water coefficient 𝐿 Τ 𝑝 𝑥 Transfer coefficient is free diffusion over the pore radius 11

Particle scale extraction – intraparticle diffusion Oil-water partitioning and diffusion Hindered diffusion Mesopore Nanoporous coffee cell wall Released aroma undergoes hindered diffusion to the particle surface Diffusivity calculated by Maxwell homogenization of free and hindered diffusion 12

Particle-water-steam transfer 𝐼 𝐷 𝑞,𝑆 𝐷 𝑥 𝐷 𝑥 𝐷 𝑕 𝑆𝑈 (1) 𝑙 𝑥 𝑙 𝑕 water-steam particle water steam boundary layer Water is considered a stagnant film Henry’s law volatility constant determines partition from water to air 13 1) Carberry, J. (1960). A Boundary-Layer Model of Fluid-Particle Mass Transfer in Fixed Beds. American Institute of Chemical Engineers Journal, 6 (3), 460-462

Column scale advection Pressur e and Source: Advection temper Water-gas mass transfer ature gradient Fresh steam enters the column base Open boundary at the column top Steam 𝑑 𝑕 𝑨 = 0, ∀𝑢 = 0 Everything that exits the column is condensed 14

Model results Comparisons with data from Mateus et al. (2007) Mateus, M.L., Lindinger, C., Gumy, J.C. and Liardon, R. (2007c). Release Kinetics of Volatile Organic Compounds from Roasted and Ground Coffee: Online Measurements by PTR-MS and Mathematical Modelling. Journal of Agricultural and Food Chemistry, 55(25), 10117-10128

Published vs Model data – acetaldehyde – the ‘simple’ aroma Single coarse particle Coarse + fine particle Optimized cell wall porosity (%) 3.3 2.8 Root mean square error (%) 6.8 5.6 16

Published vs Model data – acetic acid log 𝐼 𝑈 𝑐 = − ∆𝐼 1 − 1 𝐼 298 𝑆 𝑈 𝑐 298 Determines Henry’s constant relation with temperature Average Optimized 6400 9408 ∆𝐼 𝑆 [𝐿 −1 ] Root mean square error (%) 67 17 17

Published vs Model data – reactive aroma - Pyridine Aroma-matrix interactions are widely discussed in literature 1-3 Assuming a 1:1 ratio or reactants and a reaction rate first order in both reactants leads to a good approximation Optimized 8.1x10 -4 𝑙 𝑝𝑜 Root mean square 5.2 error (%) 18 1.Guichard, E. (2015) Interaction of aroma compounds with food matrices in Parker, J.K., Elmore, S., Methven, L. and José, M. (eds.) Flavour Development, Analysis and Perception in Food and Beverages. Cambridge: Woodhead Publishing, pp. 273-295 2.Hofmann, T., Czerny, Calligaris, S. and Schieberle, P. (2001) Model Studies on the Influence of Coffee Melanoidins on Flavor Volatiles of Coffee Beverages. Journal of Agricultural and Food Chemistry, 49, pp. 2382-2386 3.Charles-Bernard, M., Kraehenbuehl, Rytz, A. and Roberts, D.D. (2005) Interactions between Volatile and Nonvolatile Coffee Components. 1. Screening of Nonvolatile Components. Journal of Agricultural and Food Chemistry., 53 (11), pp. 4417-25

Model results Predictions of plant scale steam stripping

Extraction is determined by polarity and partitioning behaviour Furaneol concentration in the steam throughout the column with time 20

The effect of steaming time Very polar compounds (e.g. furaneol) increase in concentration with time Other compounds are diluted, with reactive compounds diluting most and apolar compounds diluting the least 21

Optimizing extraction Grind size and steaming time are significant variables but they affect aromas differently Coarse grinds and long times favour polar compounds Finer grinds and short times favour guaiacols 22

Summary and next steps

Summary Aroma steam stripping can be modelled by oil release, diffusion, • Henry’s law partition and advection Three categories of aroma can be identified by physical-chemical • properties Process variables (time and grind size) can be optimized for particular • compounds and aromatic notes Next Steps Extraction from partially wet coffee + wetting kinetics • Experimental data at lab and pilot plant scale • Identifying reactive compounds • 24

Acknowledgement and thanks

Thank you Any questions?