30% The data were analysed to look for correlations between a number of parameters (grape way all winemakers judge and plan ferments. Othersinitial may be more by peak Baume variety, type of yeast inoculation, DAP addition, SOguided 2 content) and the deviation 20% 5% drop per day somelength other factor. In future, is hoped incorporateand more targets between theor desired of the ferment asitinput by thetowinemaker theflexible actual duration. for ferment performance as options in thea ferment Fermentwas Simulator. should also be This gave the best indication of whether runningItmore quickly ornoted slowlythat than 10% 0% 0% the shows correlation, but does not necessarily demonstrate the causeisof thenecessarily differences. thedata winemaker’s ideal, though it should be noted that ferment duration not the The relationships observed are intended to provoke thought and further investigation. As Predic�on Score 30% grows, more correlations will come to light and greater confidence may 100% be had Figure 1. Prediction scores for all ferments from vintage 2017. A prediction score below 25 was classed the dataset in the observations. as a good fit between the simulation and the provided Baume data. 90% 25% 40% 80% The deviation between the desired and actual duration of all ferments was70% normally 35% 60% 15% Cumula�ve ← Good while ﬁt: 81% Poor ﬁt: 19% →more likely to be slower than chance of running faster than planned, red ferments were 50% the winemaker’s target. A breakdown by the major varieties from the dataset is shown in 40%towards Figure 3, with Cabernet Sauvignon and Merlot ferments showing greater tendency 10% 30% a role in sluggish ferments than other varieties. The timing of the ferment also seemed to play ferment performance, with ferments starting in early to mid-vintage more likely 20% to be faster Percent of Ferments Percent of Ferments 20% distributed around zero deviation. Of the ferments simulated, white ferments had a higher 5% Red 30% White 25% 20% 15% 10% than planned, and ferments beginning towards the end of vintage more likely to10% be seen slower. with a total DAP addition between 100–200 ppm, and an increase in the number of 5% rapid ferments seen with total additions >200 ppm, as might be expected (Figure 5). There 0% seen with a total DAP and anwith increase in the yeast number ofthose A difference wasaddition also seenbetween between100–200 fermentsppm, inoculated propagated andwas no difference 2-4 in performance starting with 0–10 of free>20SO 4-6 6-8between 8-10 ferments 10-12 12-14 14-16 16-18 ppm 18-20 0% 0% 2 rapid ferments seen with total >200 ppm, as might4),bewith expected (Figure 5). There inoculated with active dryadditions wine yeast (ADWY) (Figure ADWY-inoculated ferments Ferment (days) It wasdura�on not until free SO2 was >20 ppm that an compared to 10–20 ppm free SO2 (Figure 6). Predic�on Score Figure 2. Distribution of ferment duration for red and white ferments. a higher propensity between to finish ferments more quickly thanwith the 0–10 winemaker’s target, while was nohaving difference in performance starting ppm of free SO 2 influence on ferment performance became apparent. Figure 1. Prediction for allfrom ferments fromAvintage A Figure 1. Prediction scoresscores for all ferments vintage 2017. prediction2017. score below 25 was classed Figure 2. Distribution of ferment duration for red and white ferments. propagated ferments higher togood be slow. of diammonium (Figure 6).classed Itpropensity was not free SObetween was >20 ppm that an compared 10–20 ppm free SO asprediction atogood fityeast between the simulation and the provided data. 2had 2Additions score below 25a was asuntil aBaume fit the 12 simulation and the provided Baume phosphate (DAP) also had a significant effect,data. with a reduction in the number of slow ferments influence on40% ferment performance became apparent. 35% 100% Red 80% 90% 80% 25% 70% 80% 20% 60% 70% 15% 50% 60% 40% 10% 50% 30% 40% 5% 20% 30% 0% 10% 20% 0% August 2017 90% White Percent by Performance Percent by Performance of Ferments Percent Percent by Performance 100% 90% 30% Technical Review No. 229 100% 70% 60% 50% 40% 30% 20% 2-4 4-6 6-8 8-10 10-12 12-14 14-16 Ferment dura�on (days) 16-18 18-20 Chardonnay Sauv Blanc Semillon Shiraz Cab Sauv Figure 2. Distribution for red and white n=124 of ferment n=47 duration n=59 n=69 ferments.n=25 0% <-6 Good >2.5 >6 Ac�ve Dry Wine Yeast<-2.5 Propagated Yeast 12 Technical Review No. 229 n=207 n=319 10% 10% >20 0% Merlot n=39 Ac�ve Dry Wine Yeast n=207 <-6 <-2.5 Propagated Yeast n=319 Good >2.5 >6 August 2017 4. Number of days deviation from desired ferment duration, broken down by inoculation Figure Figure 3. Number days deviation from desired ferment duration, Figure 4. Number of days deviation from desired ferment duration, Figure 3. Number of daysof deviation from desired ferment duration, broken down by grape variety. strategy. Good performance was classed as finishing within ±2.5 days of the desired duration (time <-6 <-2.5 Good >2.5 >6 was classed as broken down by grape variety. Good performance broken byainoculation performance was classed as Good performance was classed as finishing within ±2.5 days of the desired duration (time to 0 Baume). to 0 Baume). Red down indicates rapid ferment,strategy. while blueGood indicates a sluggish ferment. Figure 4. Number awithin ofrapid daysferment, deviation from desired ferment duration, broken by inoculation Red indicates while indicates a sluggish ferment. finishing ±2.5 days ofblue the desired duration (time to down 0 Baume). Red finishing within ±2.5 days of the desired duration (time to 0 Baume). Red strategy.indicates Good performance classedwhile as finishing within ±2.5 a days of the desired duration (time a rapid was ferment, blue indicates sluggish ferment. indicates a rapid ferment, while blue indicates a sluggish ferment. 100% to 0 Baume). Red indicates a rapid ferment, while blue indicates a sluggish ferment. August 2017 Technical Review No. 229 13 90% 100% 90% 80% 90% 80% 70% Percent Performance Percent byby Perfromance 100% Percent by Performance 80% 70% 60% 50% 40% 30% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 20% 0% 0% 10% 0% 0-99 ppm n=395 100-200 ppm n=92 >200 ppm n=39 0-100-99 ppmppm n=257 n=395 10-20 ppm 100-200 ppm 20-30 ppm n=163 n=34 n=92 <-6 <-2.5 Free SO2 Good >2.5 30-40 >200 ppmppm n=15 n=39 >6 <-6 desired <-2.5 ferment Good >2.5 >6 Figure 5. Number of days deviation from duration, broken down by total DAP addition. Good performance was classed as finishing within ±2.5 days of the desired duration (time <-6 <-2.5 Good >2.5 >6 Figure 6. Number of days deviation from desired ferment duration, broken down by free SO2 Figure 6. Number of ferment, days deviation desired ferment duration, to 0 Baume). Red indicates a rapid while bluefrom indicates a sluggish ferment. concentration in pre-ferment juice concentration analysis. Good performance was classed finishing within ±2.5 Figure 5. Figure Number5.ofNumber days deviation from desired ferment duration, broken down by total DAP of days deviation from desired ferment duration, broken down by free SO2 in pre-ferment juiceasanalysis. days of the desired duration (time to 0 Baume). Red indicates a rapid ferment, while blue indicates addition. Good performance classed finishing within days of the desired durationas (time broken down bywas total DAPas addition. Good±2.5 performance was classed Good performance was classed as finishing within ±2.5 days of the a sluggish ferment. 14 Technical Review No. 229 August 2017 to 0 Baume). Red indicates a rapid ferment, while blue indicates a sluggish ferment. 14 finishing within ±2.5 days of the desired duration (time to 0 Baume). Red indicates a rapid ferment, while blue indicates a sluggish ferment. Technical Review No. 229 The larger the dataset becomes, the more detailed information will be available. Eventually it should be possible to break down the data by individual varieties in individual regions, with different yeast strains, etc. Australian wineries are encouraged to consider using the Simulator for vintage 2018, both for the immediate predictive benefit in the winery, and as a way of contributing to stronger fermentations for January 2018 – Issue 648 August 2017 desired duration (time to 0 Baume). Red indicates a rapid ferment, while blue indicates a sluggish ferment. Conclusions The results shown highlight some of the ways in which the data from the Ferment Simulator the whole Australian wine industry. funds from the Australian government. can be used to gain greater insights into ferment performance and potential drivers of The Ferment Simulator can be The AWRI is a member of the Wine fermentation issues. as ThealargerInnovation the dataset becomes, thein more detailed information will be accessed by all Australian wineries Cluster Adelaide, SA. available. Eventually it should be possible to break down the data by individual varieties in free module of The WineCloud at https:// individual regions, with different yeast strains, etc. Australian wineries are encouraged thewinecloud.com.au/. Tadro Abbott, Project Engineer: tadro. to consider using the Simulator for vintage 2018, both for the immediate predictive benefit in abbott@awri.com.au the winery, and as a way of contributing to stronger fermentations for the whole Australian Acknowledgements The AWRI’s communications are This article originally appeared in the wine industry. supported by Australia’s grapegrowers AWRI’s Technical Review and winemakers through investment The their Ferment Simulator can be accessed by all Australian wineries as a free module of The body, Wine Australia, with WineCloud atmatching https://thewinecloud.com.au/. www.winetitles.com.au Acknowledgements Grapegrower & Winemaker 55