Chemical Composition and Low-Temperature Fluidity Properties of Jet Fuels

Round 1

Reviewer 1 Report

Congratulations on carrying out interesting research. I have the following comments.

Major

1. Introduction

• Authors should indicate what distinguishes their research. What is the application of their research results.
• Literature is poor. Most of the publications are over 10 years old.

2. Results

•  It is worth discussing sample No. 26 in detail due to the significant difference from other results (Fig. 2)
• No discussion of the results
• No analysis of conclusions in relation to the literature

Minor

• Incorrect chapter numbering.
•  Wrong numbering of tables
•  A clearer presentation of linear regression formulas is recommended

Recommendations

• In the introduction, indicate the purpose and value of the research carried out.
• Discuss the results
• The content of the article includes indications for biofuels. What impact can the addition of biofuel have on the results of the analyzes performed? Is it possible to model mixtures?

Author Response

Thank you very much for your useful review.

Please see the attachement.

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript by Alirio Benavides et al. deals with correlation between low-temperature properties and composition of jet fuel. Although many measurements were performed (70 samples of jet fuel), presenting average values is not too useful, especially if there are also some special fuels with specific composition. In my opinion, creation of statistical models based on the content of individual components (forming very small part of the sample) is not suitable. Here are just some of my other comments.

Materials and methods:

In my opinion, better description of the samples should be provided (at least the number of samples of JET A, JET A1 and other types).

I also suggest better description of the method used for determination of composition and its procedure for quantification. Two methods, on which the method used was based on, are different. Information about column dimensions are also missing.

I do not understand “to get between 140 and 160 signals in the chromatogram”. Do authors means number of peaks, or number of selective ion (m/z) chromatograms?

Table 2:

In my opinion, presenting the mean value and standard deviation can be misleading as there are 70 different samples.

Tables 3-5:

The range of main individual components content (% wt.) should be also written in tab. 3-5 in order to prevent misleading interpretation. It can be confusing that there are the same abbreviations (C1-C10) for different compounds in Table 1 and Table 4.

Fig. 1:

It is not easy to read legend of classes designation back in Tab. 1.

Table 5, Fig. 3:

In my opinion, the model based on the content of several compounds, whose sum is so low (total wt.%) is not useful and reliable. Low R² values confirm that. Moreover, Fig. 3 presents in fact the same results as Table 5.

Table 6, Fig. 4:

Table 6 and Fig. 4 present the same results. Moreover, discussion about regression coefficients could be misleading. The statistic model was not described in detail and it is possible that different model with similar R² values can have different regression coefficients. In my opinion, it would be much more useful to present the differences between measured and calculated values.

Author Response

Thank you very much for your review.

Please see the attachement.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors took into account the suggestions and answered the questions posed in the review

Author Response

Dear Editor, 

Thank you so much for your support. Also, we want to thank for his/her revision. 

Reviewer 2 Report

The manuscript by Alirio Benavides et al. was revised and improved. In spite of that, it should be carefully checked again. Here are my suggestions and comments:

Tables 3-5

I appreciate addition of mean values of the components, but I suggest present rather the range (minimum and maximum values). Standard deviation can be misleading, because it is calculated for different fuels. It can be well seen in table 2. The range of paraffinic compounds content is from 57 to 75 %, but standard deviation is “only” 4 %.

Fig. 1:

Authors removed the same designation of some different hydrocarbon groups and compounds, but now, the designation in the legend do not correspond to any hydrocarbon groups.   

Table 6, Fig. 3 (no reaction)

In my opinion, the model based on the content of several compounds, whose sum is so low (total wt.%) is not useful and reliable. Low R values confirm that.

Table 6, Fig. 4:

Table 6 and Fig. 4 present the same results. I suggest to describe the statistic model in more detail.

Author Response

Comments and Suggestions for Authors

The manuscript by Alirio Benavides et al. was revised and improved. In spite of that, it should be

carefully checked again. Here are my suggestions and comments:

Tables 3-5

I appreciate addition of mean values of the components, but I suggest present rather the range

(minimum and maximum values). Standard deviation can be misleading, because it is calculated

for different fuels. It can be well seen in table 2. The range of paraffinic compounds content is

from 57 to 75 %, but standard deviation is “only” 4 %.

As suggested by the reviewer, the standard deviation is not considered in new tables 2-5 (a new

column with the maximum value was added). It is worth to mention that the mean value was

calculated taking into account the whole sample of fuels (70). It is possible that some main individual components (those present in at least 50% of the samples and/or with a weight percent greater tan 35%) do not appear in all fuel samples.

Fig. 1:

Authors removed the same designation of some different hydrocarbon groups and compounds,

but now, the designation in the legend do not correspond to any hydrocarbon groups.

To avoid any confusion, Figure was changed in total agreement with the notation used in table 1.

Table 6, Fig. 3 (no reaction)

In my opinion, the model based on the content of several compounds, whose sum is so low

(total wt.%) is not useful and reliable. Low R values confirm that.

In our opinion, it is particularly useful to present the information in both formats, tables, and figures. Although, the information is similar, in the figures only the components with strong influence are presented.

In my opinion, the model based on the content of several compounds, whose sum is so low

(total wt.%) is not useful and reliable. Low R values confirm that.

The reviewer is wright. In the manuscript the reason for the low R values obtained by using the first compositional approach is discussed and then the second compositional approach is presented, and higher R values are obtained.

Table 6, Fig. 4:

Table 6 and Fig. 4 present the same results. I suggest to describe the statistic model in more detail.

In our opinion the static model is presented with sufficient detail and we think that the manuscript

will be too long by adding more details.

The following paragraph is part of the manuscript:

“To evaluate the composition-properties relationship, a statistical analysis was performed selecting composition as input variable, and fluidity properties as output or response variables. To identify patterns in the analyzed data, a correlation matrix was calculated. Regarding the output variables, it was determined that all the variables had significant linear relationships, direct (positive coefficient) or inverse (negative coefficient). In the case of the input variables, most of them (26) had significant direct or inverse linear relationships. Given the strong correlation between the variables, dimension reduction methods were applied to obtain variables that did not violate the assumption of non-existence of multicollinearity of the regression. The dimension reduction method chosen was principal components, which is a mathematical procedure that transforms a set of correlated variables into a set of new uncorrelated variables to reduce the size of the data and facilitate its interpretation and analysis [22]. Following this approach, regressions of each response variable against the main components and hydrocarbon families as input variables were obtained with a confidence level of 90% and p values < 0.1. It is important to highlight that all regressions were validated according to the required assumptions of no multicollinearity in the input variables, homoscedasticity, no autocorrelation and normality of the residuals”.