M-8: Yeast production and multivariate statistical process control: Utilization of on-line data for statistical control of yeast propagation as a batch-wise process

I. J. WHITEHEAD (1), D. Krause (1), M. A. Hussein (1), T. Becker (1); (1) TU München Lehrstuhl für Brau- und Getränktechnologie, Freising, Germany

Thursday, June 5 - 4:00 p.m.-5:45 p.m.
Level 4, Grand Ballroom

Controlling and sustaining a standard yeast production process is essential for maintaining consistent quality assurance in beer production. Traditionally, yeast production has been controlled by off-line laboratory evaluations of yeast counts, infections, vitality, and viability. These methods are extremely time-consuming and costly, require trained personnel, and may only be performed at distinct time intervals. An alternative method of analyzing and controlling yeast production in which on-line and off-line data pools are used to create soft sensors that can determine several critical process parameters is presented. The presentation shows how such sensors can be applied to the development of a multivariate statistical process control (MSPC) system that can be used to continuously monitor and control yeast production in real time. On-line sensors can be used to collect large amounts of data in the yeast production process. While various MSPC methods have been described and implemented in a number of production settings, the development of a MSPC system for yeast production in a brewing setting is complicated by the substantial variations in batch lengths. However, a more recent MSPC approach has been applied to allow for large differences in batch lengths. This approach prescribes a method of monitoring the whole batch postproduction and its evolution in real time. An on-line array of sensors has been implemented, and the data generated by these sensors have been combined with off-line laboratory results. Multivariate calibration techniques have been used to analyze this data and develop a range of software sensors that can accurately monitor several critical process parameters in the yeast production process in real time. These software sensors together with the MSPC program can be used for the on-line monitoring and control of yeast production. To demonstrate the feasibility of this approach in a production setting, a real-time control system was successfully implemented on a 60-hL pilot bioreactor at the TUM Weihenstephan Department of Brewing and Beverage Technology. A soft sensor that is able to predict yeast cell counts with an accuracy of 7 million cells/mL combined with a MSPC system is currently under development. The advantage of adopting such a system is that it allows real-time feedback of process parameters that would normally be time-consuming lab tests, thus providing faster reaction to production problems. Control of variations within the production process through MSPC allows for more consistent yeast production, as well as identification of inconsistencies in production causing fluctuations in the quality of yeast produced. The approach described in this presentation could also be applied to other aspects of the production process in the brewing industry.

Iain Whitehead began studying mathematics at the Memorial University of Newfoundland, Canada. His interest in brewing brought him to Germany, where he completed an internship at the Spital Brewery in Regensburg, Germany. After registering for the diplom braumeister program at TUM Weihenstephan, Iain completed a 12 month internship at the Augustiner Brewery in Munich, Germany. After his successful completion of the diplom braumeister program in September 2012, Iain accepted a position at the TUM Weihenstephan Chair of Brewing Beverage Technology and is currently involved in several of TUM Weihenstephan’s ongoing research projects.

Brewing Resources

Ask the Brewmasters Technical Quarterly MBAA Webinars MBAA Podcasts Food Safety Brewery Safety Vendor Search