O-22
Comparing flavor of Japanese beer using sensory and instrumental analyses
Presenter: Kazutaka Ozaki, Asahi Breweries, Ltd., Moriya, Ibaraki, Japan
Co-Authors: Minoru Kobayashi and Hiroshi Yamashita, Asahi Breweries, Ltd., Moriya, Ibaraki, Japan; and Tetsuo Aishima, Chemometrics & Sensometrics Laboratory, Saitama, Japan

Since the late 19th century, lager-type beer has mainly been consumed in Japan. As widely recognized, the difference in sensory properties among beers belonging to the same category is rather small. In this research, we attempted to compare sensory properties in major brands of beer currently distributed in Japan and interpret these differences in terms of chemical components so as to aid designing and developing new products that can meet the preference of consumers. The sensory properties of 10 beers composed of 6 ordinary lager-type beers and 4 low-malt beers, the market share occupied by these 10 brands attaining nearly 99%, were analyzed by quantitative descriptive sensory analysis using a well-trained panel. Nonvolatile taste components were analyzed by HPLC and routine chemical analyses. After extracting volatile components in beer with dichloromethane, the aroma extract was analyzed by GC-MS. The CHARM value or magnitude of aroma note of each volatile component was determined by GC-O, which uses human olfaction as a detector, of a dilution series of the original aroma extract. The data sets thus obtained were analyzed by principal component analysis (PCA). The partial least squares regression analysis (PLSR) was applied to sensory and chemical data sets to calculate PLSR models that can predict each sensory attribute. The attributes identified by sensory analysis were 28: 13 for aroma, 7 for taste, 4 for flavor, 1 for aftertaste, 2 for mouthfeel and total thickness or body. According to instrumental and chemical analyses, more than 200 volatile and nonvolatile components were identified and quantified. Among them, aroma compounds determined were about 150 in GC-MS and 80 in GC-O. The PCA of sensory data revealed the presence of two groups each corresponding to ordinary lager-type beers and low-malt beers, and other sample clusters according to manufactures were also observed The PCA of whole chemical components showed the sample clustering according to manufactures. However, the PCA of volatile components indicated clear separation of ordinary lager-type beers and low-malt beers. Highly predictive PLSR models were obtained for each of taste attributes, such as bitterness, sweetness, sourness, astringency, acridity, cereal taste, and aftertaste, by selecting highly contributing nonvolatile components. PLSR models obtained for individual aroma attributes by selecting aroma components according to their statistical contribution to each attribute were highly predictive. Considering the taste or aroma notes of chemical components selected by PLSR as important for each attribute, results of PLSR were mostly acceptable. Thus, integration of sensory and chemical data sets using multivariate analyses can supply information that can play key roles in understanding sensory properties of Japanese beers.

Kazutaka Ozaki graduated from Chiba University with a B.Sc. in chemistry. He joined Asahi Breweries, Ltd. in April 1990. He has spent 4 years at the Nagoya plant as staff of the quality control section. In 1994, he transferred to the Brewing R&D Laboratory of Asahi Breweries. He has been engaged in research analytical technology and sensory evaluation. He also has served the analytical subcommittees of BCOJ from 1996 to 2002. He is currently the Manager at Analytical Technology Laboratory of Asahi Breweries, Ltd.