A Model of Catalytic Cracking: Product Distribution and Catalyst Deactivation Depending on Saturates, Aromatics and Resins Content in Feed
Round 1
Reviewer 1 Report
The authors have proposed a new model -- kinetic scheme of catalytic cracking to predict the contents of the gasoline and the gas hydrocarbons -- called "riser model" and a set of validational experiments. This appears to be novel and original, so eventually a quality journal article should result.
The major problem of the article is that the lack of clarity violates the first principle of peer review. Peers should be able to reproduce the results of the paper, so sufficient information should be provided for peers to do so. At the heart of this paper is an 18-concentration system of first order, ordinary differential equations in "space-time" (the coordinate along the axis of a plug flow reactor) and a thermal energy equation of the same class.
I defy any expert in reaction networks to produce, from the information given, the terms in equation (1) so that they could actually numerically integrate the system of equations of the riser model. The information just is not given, for instance, how the W-arrow terms depend on concentration of the 18-concentration vector. The values of some kinetics coefficients are given in Table 3. How do those relate to the unstated kinetics functions W-arrows? How can this information possibly be extracted from the DFT diagram of the kinetics network in Figure 3.
Some experimental data is presented in subsequent graphs that are bar charts for percentage changes and compared to model predictions. Bar charts are a dreadful way to compare performance -- worse than jus providing comparison tables. In the Methods section, the reader is told that inverse kinetic methods are used, but this is not explained. No references are given. From experience with data assimilation, the reader needs to be told what data was used, and the algorithms employed, to infer the unknown parameters. Typically, datasets are split into "training data" for assimiliation, and validation data, for comparison of the predictive, i.e. semi-empirical representation of the experimental data. Graphs / tables for how good the data assimilation represent the training data and separately the predictive value need to be clearly demarcated.
Overall, this paper is an exercise of the authors talking to themselves, not the audience. Perhaps the authors in a revision can imagine themselves in the position of the audience, and ask "what do I need to know in order to use this model?" as well as what are important ways to benchmark it?
Separately from the presentation of the work, there is the typical matter of presenting how well the numerical analysis can be shown as a faithful approximation of the model equations themselves. For a set of ordinary differential equations that are first order, historically one would use Runge-Kutta methods and show convergence with time stepping reduced. A discussion of whether or not the equations are stiff also is important. Nearly irreversible reactions make for stiff systems, which can now easily be integrated using Gear's method for backward / implicit time stepping locally (built into Matlab, for instance). Since there is no discussion of the fidelity of the numerical analysis, the authors may be interpreting "garbage out". People are aware of "garbage in, garbage out" with numerical analysis, but with stiff systems, it can be "proper information in, garbage out" due to the pathology that stiff system approximations get worse with time step reduction.
I am afraid that this is an interesting study with an incomplete primary disclosure, so must be pro forma rejected.
Author Response
Response is attached.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper describes a numerical model to optimize refinery parameters, in order to increase yield and quality of the fuel produced, and minimize catalyst deactivation.
The subject is surely relevant for industrial production, but the actual manuscript seems to be a variation of other very similar work, and therefore of limited novelty.
For example, in ref 19 and 20, by the same author , very similar models are proposed, and the “mass spectrometry and chromatographic data” are not reported.
Moreover, the quality and the English of the paper must be substantially improved.
For example:
Abstract, line 10: coke is unavoidable, it is not completely true that it is due to residuals
Line 12, it would be better to substitute with “ in order to minimize the catalyst deactivation”
Line 13 very poor English
Line 40-42 delete the bibliographic data in the text (they are correctly listed in the reference section)
Line 162 it is not clear how quantum chemical calculations (DFT) were used
Table 3, what is the meaning of the asterisk?
Figure 4 Some legend is missing , why two sets of experimental and calculated data are shown?
Line 255-256 Very poor English
Author Response
Response is attached
Author Response File: Author Response.pdf
Reviewer 3 Report
The main aim of this paper is to investigate the mathematical model of catalytic cracking process, taking into account the kinetics of the compounds’ group as well as the activity change of catalysts. The feedstock contained saturated and aromatic hydrocarbons and resign; its exact composition is not given. The cracking reaction is described by 18 reversible reactions applying simple diffusion kinetics, assuming a plug flow reactor (but the kinetic equation does not involve the convective flow?). On the other hand, the potential role of the dispersion, and accordingly values of Peclet number are also recommended to be discussed shortly. The reaction rate expressions are also not discussed, though the reaction rate constants are predicted (Table 3).
Generally, the whole material is not suitably compact, not precisely formulated, the text is hard to understand. In the most cases, regarding e.g. tables and figures, the brief explanations are often missing; an abbreviation list is also needed (there are not defined e.g. the meaning of m, z, CNAC, RON). How the effect of the concentrations are taken into account in the reaction rate expressions [see Eq. (1)]? Authors apply plug flow model (line 177), then where is the effect of the convective velocity (it is missing in the kinetic equation)? deltaH and deltaG change in a given range (Table 3), how the effect of mixed feedstock was taken into account (this needs brief explanation in the text). Conclusion should be more concrete and more precisely formulated, emphasizing the more important results of the paper. On the other hand, results of the experiments and simulations are useful and they can be regarded as valuable ones, thus the manuscript can be acceptable after major revision. Some minor remarks to be taken into account in the revised version:
- Caption of figures, tables should be much more informative; they must involve the important circumstances of the data presented.
- Line 67: what is g/sm3?!
- Line 135: remove Russian letter!
- Line 136: 4.1/2.3 (Table 1) is equal to 2.1?
- Table 2: what does mean S catalyst?
- Line 153: z/m=44 (in Fig. 2 is m/z; which one is correct?); what does z/m mean?
- 2: please identify the curves on this figure;
- 3: How can be the same the reaction rate constant of two reactions: e.g. k1 for alkanes, k1 for unsaturated HC?
- Table 3: are the reaction rate constant really constant at different mixed feedstocks? What can be the error of this assumption, really less than 1%? What kind of unit of measure is mol ls/s, which follows from s-1/ls-1mol-1? What is ls?
- Table 4: kgf/cm2; it is more correct to be used Pa, or bar! What material is the sludge here?
- Fig 4: Why are two experimental and model results plotted? There is no any explanation for it in the text;
I recommend a careful rewriting of the material with more precise formulation of the methods, the experimental conditions, the experimental results and also the theoretical ones with deeper information on the mathematical model, and more clearly formulation of the conclusion, with more efforts on the emphases of the essence of results.
Author Response
Response is attached
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
'invers' is misspelt in line 173.
The authors have made substantial modification to the manuscript to address the major communications failings in the original version. They also provided substantial information in the response to my comments about the clarification of the inverse method used -- by genetic algorithms -- and the variation with residence time step. A reference 16 which is not in the paper to genetic algorithms, subsequent references 17-32 about modelling, and a graph of sensitivity of one component to time step are presented. None of this information are actually included in the paper, as per my original report.
I know perfectly well how to implement multiple different inverse methods, but that does not mean that telling the referee is sufficient. It must be explicitly stated in the paper. Whatever these references 17-32 are that are not in the paper, it you should actually explain in the paper what there value is, not just to the referee. They do not seem directly connected to the response to comment 3 -- no citations -- so please explain how they connect to the material that is in your response to comment 3, but is not put in the paper revision. If it answers my questions, it should be in the revision, as I am representative of your target audience. The barrage of references is meaningless unless you explain its significance. Not to me, but to the audience of the paper!
Finally, the graph of sensitivity to step size is worrisome (and not included in the paper). The authors have the responsibility to show convergence with the step size used. This graph does not show convergence with step size h=0.01 -- it shows a large divergence from much larger step sizes. Hence it still begs the question, "How do you know you have converged sufficiently?" That question needs to be answered. If you have, then the issues surrounding stiffness of the system are either answered or exposed.
With all the work that the authors have done to improve the clarity, it still falls short on these points. Hence there is still more work, as listed above, to complete before the manuscript is of publication quality. This should not take long, but please learn the lesson from this exchange that you address all points that you can on clarity within the revision. Perhaps referees have misunderstood what you pose, but if they have done so, diagnose from their comments what the source of their misunderstanding is, and make it clearer. Ultimately, publication is about clear communications. Most of life's errors stem from miscommunications!
Author Response
Response to reviewer is attached.
Author Response File: Author Response.pdf
Reviewer 3 Report
Authors have revised their manuscript with increasing its scientific level. I have only one insufficiency: the abbreviation list is missing, though it would be important for the better, quicker understanding, to my opinion.
Author Response
Response to reviewer is attached.
Author Response File: Author Response.pdf
