Seeded Search Estimation Algorithm for On-Line Prediction of Wine Fermentation
James Nelson,* Robert Coleman, and Roger
Boulton
*University of California, Davis, 1 Shields Ave, Davis, CA, 95616
(jjnel@ucdavis.edu)
Mathematical models for wine fermentations and parameter estimation algorithms have been used to analyze wine fermentations; however, such analysis is performed off-line, typically after fermentation is completed. Combined with automated Brix measurements, an on-line model estimation system would allow automated prediction of fermentation density and quantification of rates of energy and CO2 evolution across all concurrent fermentations in a winery. The overlapping curves of many concurrent fermentations – with different starting times, temperatures, volumes, starting compositions, inoculations, etc. –yields highly non-uniform loads. The electricity charges for such loads are based on the maximum peak rate of use (kW), the energy used (kWh) and the time of use during each day. The measured and predicted rates of fermentation energy can be used to manage refrigeration loads and their energy requirements. Additionally, the rates of CO2 evolution can be used to manage loads on future carbon capture and sequestration systems. Here, an on-line seeded search parameter estimation algorithm, in combination with the Boulton fermentation model, is proposed to predict fermentation trajectory and provide a confidence in the estimate. The method was evaluated on a set of red and white fermentations. From the rate of fermentation, the rates of energy and CO2 production were also compared to the on-line model predictions. As more data becomes available, confidence in the on-line predictions increases and closely follows the automated measurements. Finally, the summation of overlapping energy and CO2 loads are calculated from automated Brix measurements for all fermentations from the 2021 harvest at the UC Davis Teaching and Research Winery.
Funding Support: T.J. Rodgers Fellowship in Electrical and Computer Engineering, Treasury Wine Estates, Stephen Sinclair Scott Endowment in Viticulture and Enology
