Methods for Parameter Estimation in Wine Fermentation Models
Guojie Jin
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
A very interesting article on the development of computational tools for monitoring alcoholic fermentation based on laboratory measurements of parameters that change during alcoholic fermentation.
However, some questions need to be answered and some issues need to be clarified before the article can proceed for publication.
General comment.
The study of biochemical processes in general is quite difficult to be simulated by mathematical models unless many variables and how they interact with each other are taken into account. In the case of alcoholic fermentation, factors such as the population of yeasts starting the process, the presence of oxygen in the early stages of the process for yeast proliferation, alcohol production, temperature and fermentation with more than one strain are among the variables to be considered. The amount of sulphur dioxide is also an important factor affecting the action of the yeasts, particularly during the initial stage of vinification and during the lag phase. These parameters should be taken into account in the approximations made when solving the equations.
Specific comments
1. What is the significance of the negative values for Brix in Figure 1 and Table 1 of the supplement?
2. What was the initial yeast population at the start of alcoholic fermentation?
3. What is the physical meaning of the specific maintenance rate?
4. Why are the subsequent phases of the yeast life cycle not discussed? Do they not affect alcoholic fermentation?
Author Response
Comment: General comment.
Thank you for your comment. The study of biochemical processes in general is quite difficult to be simulated by mathematical models unless many variables and how they interact with each other are taken into account. In the case of alcoholic fermentation, factors such as the population of yeasts starting the process, the presence of oxygen in the early stages of the process for yeast proliferation, alcohol production, temperature and fermentation with more than one strain are among the variables to be considered. The amount of sulphur dioxide is also an important factor affecting the action of the yeasts, particularly during the initial stage of vinification and during the lag phase. These parameters should be taken into account in the approximations made when solving the equations.
Response: Thank you for your comment. The added sulfur dioxide to a juice ends up in various forms, approximately half is bound to sugars, more is bound to anthocyanins release in contact prior to fermentation in the case of red wines, and the free form that can affect yeast shows up in an extended lag period. Since the model estimates the lag period as one of the model parameters, any effect of yeast activity on this due to whatever the sulfur dioxide concentration is, is captured in the lag time estimate. There is no known effect of sulfur dioxide on the growth rate of yeast under wine conditions. As this study has shown the Boulton model is robust to model across yeast strain, grape varietal, temperature, initial sugar concentrations (thus final ethanol concentrations). There is a very detailed description of the model in the recent paper “Models for Wine Fermentation and Their Suitability for Commercial Applications”, Fermentation, 2024, 10, 269. https://doi.org/10.3390/fermentation10060269. This is the most comprehensive and successful model for wine fermentations today and has been applied to hundreds of commercial fermentations.
Specific comments
Comment 1: What is the significance of the negative values for Brix in Figure 1 and Table 1 of the supplement?
Response 1: Thank you for your question. Please see the calculations provided in Section 1.6 of “Models for Wine Fermentation and Their Suitability for Commercial Applications,” Fermentation, 2024, 10, 269. https://doi.org/10.3390/fermentation10060269. The density scales used in wine and beer fermentations (Brix, Baume, Oecshle and Plato) are based on the specific gravity and all range from zero at the density of water. For solutions less dense than water, these scales all have negative values. In wine at the end of fermentation, an ethanol content of 14% v/v ethanol, 6 g/L glycerol, and 25 g/L extract, the density will be 0.99, or about -2.5 Brix.
Comment 2: What was the initial yeast population at the start of alcoholic fermentation?
Response 2: Thank you for your comment. The initial yeast population was provided as initial dry active yeast concentration (g/L) at inoculation – please see Table 1, column “Inoculum”. It is different for each fermentation, and it is an initial condition of the model for each fermentation.
Comment 3: What is the physical meaning of the specific maintenance rate?
Response 3: Thank you for your inquiry. The meaning of specific maintenance rate is discussed in section 1.2 of “Models for Wine Fermentation and Their Suitability for Commercial Applications,” Fermentation, 2024, 10, 269. https://doi.org/10.3390/fermentation10060269. Here the authors state that specific maintenance rate is the consumption of sugar for the maintenance of viable biomass. It is sometime referred to as the rate of sugar consumed to account for the turnover energy of the cells.
Comment 4: Why are the subsequent phases of the yeast life cycle not discussed? Do they not affect alcoholic fermentation?
Response 4: Thank you for your question. The Boulton model includes description (differential equations) for the growth, maintenance and death of the yeast population. The cell division phase and internal rearrangements prior to that are generally lumped into the specific growth rate of all fermentation models. Such models are expressed in terms of cell mass since this is the measured variable in most fermentation studies. We refer again to the recently published research article: “Models for Wine Fermentation and Their Suitability for Commercial Applications,” Fermentation, 2024, 10, 269. https://doi.org/10.3390/fermentation10060269.
Reviewer 2 Report
Comments and Suggestions for Authors
The author of manuscript Fermentation-3089039 provides six methods for Parameter Estimation in Wine Fermentation Models. This work is of great significance to monitor, model and control the wine fermentation process, but there are some points that need to be fixed.
It is mentioned in lines 116-117 that the %RAE showed a 5.61% error in the 300 minute step size, therefore one can use up to at least 120 min, and that will result in acceptable outcomes. The upper limit of acceptable %RAE is not mentioned here, and in the corresponding table 2, the %RAE of 120 minutes reaches 2.16%, which is also very high compared with the previous several times. The acceptable range needs to be given here to draw this conclusion.
It is mentioned in lines 170-172 that the magnitude of error between parameter estimation methods was less than the magnitude of error between fermentations. The discrepancy between fermentations are not listed here. This data needs to be supplemented.
In the discussion section, the limitations or shortcomings of these models need to be given, as well as strategies for future improvements.
Author Response
Comment 1: The author of manuscript Fermentation-3089039 provides six methods for Parameter Estimation in Wine Fermentation Models. This work is of great significance to monitor, model and control the wine fermentation process, but there are some points that need to be fixed.
It is mentioned in lines 116-117 that the %RAE showed a 5.61% error in the 300 minute step size, therefore one can use up to at least 120 min, and that will result in acceptable outcomes. The upper limit of acceptable %RAE is not mentioned here, and in the corresponding table 2, the %RAE of 120 minutes reaches 2.16%, which is also very high compared with the previous several times. The acceptable range needs to be given here to draw this conclusion.
Response 1: Thank you for bringing this to our attention. We have revised the text to read “The %RAE showed a 5.61% error in the 300 minute step size, therefore if a 5% error cut off is used, one can use up to at least 120 min with acceptable outcomes.” We feel this conveys a more definitive level of error that is acceptable.
Comment 2: It is mentioned in lines 170-172 that the magnitude of error between parameter estimation methods was less than the magnitude of error between fermentations. The discrepancy between fermentations are not listed here. This data needs to be supplemented.
Response 2: Thank you for highlighting this. As a result, we have added the ‘mean’ row to Table 4 to summarize/highlight the RMSE differences between fermentations. We believe it is more clearly conveyed that, for example, the difference between fermentation 571 with an average RMSE of 0.098 and fermentation 486 with an average RMSE of 0.522 are greater than the parameter estimation method RMSE values within a single fermentation.
Comment 3: In the discussion section, the limitations or shortcomings of these models need to be given, as well as strategies for future improvements.
Response 3: Thank you for your comment. Based on the use of the word ‘model’ in this recommendation/comment, we take that to mean the fermentation model, not the parameter estimation methods. If this is the case, we would refer to the recently published work by two of the contributing authors: “Models for Wine Fermentation and Their Suitability for Commercial Applications,” Fermentation, 2024, 10, 269. https://doi.org/10.3390/fermentation10060269. This cited work provides an in-depth look into three different fermentation models. The limitations of existing models are discussed there and future directions for improving the mathematical description of wine fermentations are also given there. We feel this article answers the reviewer’s question and is appropriate for that piece of work.
Reviewer 3 Report
Comments and Suggestions for Authors
Dear Authors,
This is very interesting study which is dealing with investigation of parameters which are important in wine fermentation model. Manuscript is well written and organized. Findings from this manuscript could be applied in practice, but some information in manuscript are missing.
In the abstract are missing values of the most important results.
In the subsection 2 are missing information regarding grapes used for wine production.
Insert data regarding geographical origin of grapes, coordinates of vineyards.
On what kind of soil were grown grapes used for the production of wines?
Did you grow different grape varieties on the same soil?
What was climate conditions during these 3 different years?
Did you use any fertilizers in vineyard?
Did you use any substances against pests?
Did you check different grape cultivars for the presence of Botrytis cinerea?
What was cultivars of yeasts used for the production of commercial wines in the Table 1?
What was final alcohol content in the wines from Table 1?
Make conclusion section more direct. Highlight the most important findings from the manuscript.
Author Response
Comment 1: This is very interesting study which is dealing with investigation of parameters which are important in wine fermentation model. Manuscript is well written and organized. Findings from this manuscript could be applied in practice, but some information in manuscript are missing.
In the abstract are missing values of the most important results.
Response 1: Thank you for your comment. The most important results are that the methods are not significantly different from each other in the values that they deliver while the model can differentiate each of the fermentations from each other. This is clearly stated in both the Abstract and the Conclusion. There are no missing values in these results.
Comment 2: In the subsection 2 are missing information regarding grapes used for wine production.
Response 2: Thank you for your remark. The grape cultivars are listed in Table 1 – column “Cultivar”
Comment 3: Insert data regarding geographical origin of grapes, coordinates of vineyards.
Response 3: Thank you for your inquiry. This information has no known relationship to any of the fermentation curves analyzed or the Model parameters estimated by the alternative search methods. It would be interesting information but it adds nothing to the results and conclusions.
Comment 4: On what kind of soil were grown grapes used for the production of wines?
Response 4: Thank you for your question. This is also interesting information but there is no known relationship between soils type and site properties on the model parameters and the search methods which are the subject of this paper.
Comment 5: Did you grow different grape varieties on the same soil?
Response 5: Thank you for your inquiry. The grapes for these fermentations are from commercial vineyards in the Napa Valley. They are grown on different sites but there is no relationship between these features and any known fermentation outcomes.
Comment 6: What was climate conditions during these 3 different years?
Response 6: Thank you for your question. The climate conditions are always different from different seasons. There is no known relationship between any climate conditions and the nature of these fermentations or the parameters obtained, or of the success of the search methods employed.
Comment 7: Did you use any fertilizers in vineyard?
Response 7: Thank you for your inquiry. The vineyards from which these grapes were harvested are among the most effectively managed in California. In all of these vineyard sites, fertilization would be used when needed. These are among the most valuable grapes grown in California. Again there is no known relationship between fertilization and the fermentation characteristics or parameter estimates or search methods which are the subject of this paper.
Comment 8: Did you use any substances against pests?
Response 8: Thank you for your question. See previous answer. Again there is no known relationship with any fermentation properties due to any substances applied and the characteristics of these fermentations.
Comment 9: Did you check different grape cultivars for the presence of Botrytis cinerea?
Response 9: Thank you for your inquiry. These fermentations are from high quality grapes would have been inspected for any evidence of molds or Botrytis and not included in this project.
Comment 10: What was cultivars of yeasts used for the production of commercial wines in the Table 1?
Response 10: Thank you for your question. The yeast strains used are given for each fermentation in Table 1. They are all commercial strains sold for winemaking.
Comment 11: What was final alcohol content in the wines from Table 1?
Response 11: Thank you for your inquiry. The final alcohol content is not related to any of the model parameters in this study and does not affect any of the conclusions drawn regarding the effectiveness of the search methods.
Comment 12: Make conclusion section more direct. Highlight the most important findings from the manuscript.
Response 12: Thank you for your comment. The Conclusion section is very direct and precise in our opinion. We see no way to improve on that. The conclusions answer the objective of the paper, that is to compare search methods used in the modeling of this diverse group of commercial fermentations.
Round 2
Reviewer 1 Report
Comments and Suggestions for Authors
All queries have been respond by the authors.
I suggest the acceptance of the manuscript in the present form
Reviewer 2 Report
Comments and Suggestions for Authors
The manuscript is fine. I have no questions.
Reviewer 3 Report
Comments and Suggestions for Authors
/
