Technical Route to Achieve Ultra-Low Emission of Nitrogen Oxides with Predictive Model of Nitrogen Oxide Background Concentration
Round 1
Reviewer 1 Report
The paper presents a linear correlation of some process variables with the NOx concentration before corrective measures (SNCR) in a cement plant. One of the objectives of the study is to reduce the overdose of reductant and its associated negative effects.
The paper is generally well written and logically structured. I appreciate the use of a model that is intended to understand the process (the important function of models) instead of the trendy black box choice. The justification to build the model is that continuous measurement of NOx concentration by analyzers in the calciner is not feasible in practice due to high temperatures and dust, what is true (but there are analyzers in cement plants...). The proposed alternative is to estimate the concentration from the values of other related measured variables (a "soft sensor").
The selection of a simple formalism is an understandable and practical approach, but I doubt the model presented in the paper can correctly predict the intended NOx concentration:
1) Data regression comprises a little more than one and a half hour of operation. This is a too small time window to capture the response of a plant with high inertia and delays.
2) There is little variation of the NOx concentration in that period, the plant seems to be in a stable estate. This is fine for a linear model that can represent small fluctuations around a steady point, but unable to do the same when moving apart enough from that point. The authors indirectly mention that problem (line 331, "when the operation parameters of the grate cooler are fixed...").
3) Or consider the case of the motor current, related to the kiln torque. It is correlated with high temperature in the kiln and consequently is supposed to predict more thermal NOx production. But, unexpectedly, the resulting coefficient is negative, what, after a "spot investigation", is found to be connected to a reducing atmosphere in the kiln because an excess of coal (without an excess of air). So high current and high temperature could mean both more and less NOx... Not to mention the confusion that a ring in the kiln would create to the system.
4) It is not a surprise that the model could predict reasonably well the output of the plant in the hour immediately following the time when the data used in the regression were collected. Actually, prediction does not seem as good as declared. In figure 3 can be seen intervals (e.g. 2000-3000 s) where the forecast and the measurement increase and decrease in different senses. And in figure 6, the MAE looks higher than the 6.3 ppm mentioned in the test, but it could be a visual effect. Anyway, the fluctuation of the error is in the same order of magnitude that the fluctuation of the signal.
Generation of NOx is complex, composed of several mechanisms and occurring in several locations. It is hard to believe that a simplistic approach could be able to represent them. From this point of view, the linear model of the paper is just a nice regression exercise but without any new practical or theoretical contributions and then not yet appropriate to be published in the journal.
(Some parts of the text are not clear, like the parameters considered to select the filter time window, but lines 310-315 definitely need editing...)
Author Response
Thanks for the affirmation of the topic of the paper.
There are analyzers that can be used to carry out continuous measurement of NOx concentration in cement plant, and we believe the proposed alternative in this paper is still necessary and important for the high maintenance cost and short service life of these analyzers due to high temperatures and dust in the calciner.
1) Time window in this paper is one and a half hour, but the actual measurement time in the research process is much longer. Considering that the data in different time show similar simulation results, this paper only selects a part of them.
2) NOx concentration in the plant was in a stable estate, more testing and application will be carried out in different production lines, which will be introduced in the following articles.
3) The results and analysis in this paper were based on the real data of the production line, different production lines may show different results, and some of the results may be slightly different from existing experience.
4) Prediction model in this paper is based on linear regression method, we have reached a cooperation intention with a production line, and the application effect will be introduced in the following article.
Reviewer 2 Report
This paper deals with air pollution with NOx because of the cement production process. A lot of research has been done on this subject either by Selective Catalytic Reduction (SCR) catalysts or with non-catalytic processes. This is an important issue mainly for earth areas which are growing up fast. I think the work is well done and well presented and the results are usuful.
Author Response
The authors are grateful for the helpful recognition made by this reviewer and ongoing research will be carried out.
ongoing [ˈänˌɡōiNG] 名词前进; 行为; 举止 形容词经常的; 不断的; 前进的 定义 形容词continuing; still in progress. 同义词 形容词continuous; continuing; uninterrupted; unbroken; nonstop; round-the-clock; incessant; unending; constant; ceaseless; unceasing; endless; never-ending; everlasting; eternal; perpetual; unremitting; relentless; unfaltering 示例 ongoing negotiations
Reviewer 3 Report
I’ve revised the manuscript by Yanfei Yao and Yanxin Chen. It needs, in my opinion, of minor revisions to clarify the progress of the research respect to the current literature and to better explain the preliminary plan adopted.
Respect to the first point, I think that in the introduction section should be better explained the current state of the art.
In my opinion, the text of the publication should be verified by a professional translator.
The publication also lacks comprehensive information on measurement uncertainty and factors that limit the application of the described test method.
To sum up, I rate the article very positively and after minor corrections should be published.
Author Response
The authors are grateful for the helpful suggestions made by this reviewer.
1) Current state of Denitration technology have been supplemented, introduction of biomass denitrification and its application in cement industry are added. The relevant literature was also added and the document numbers were rearranged.
2) Factors that limit the application of the described test method have been supplemented as line 370-374.
For different production lines, the production parameters affecting the background concentration of NOx and their influence rules may be slightly different, which means that the conclusions obtained in this paper are not applicable to all production lines. In the popularization and application of the model, the above analysis process must be repeated to ensure the reliability of the results.
Reviewer 4 Report
This paper addresses the problem of the control of NOx in coal-fired cement kiln operations using selective non-catalytic reduction (SNCR) technology. The paper recognizes that SNCR is limited to a maximum NOx reduction efficiency of around 60% and it also recognizes the additional constraint of avoiding ammonia slip. The success of SNCR thus depends on the ability to reduce the uncontrolled NOx concentration at the SNCR agent injection point such that the typical 60% SNCR reduction meets the regulatory requirements. The approach is to develop a predictive model that calculates the expected uncontrolled NOx emissions as a function of measurable input parameters (e.g., coal flow, air flow via fan speed and pressure, etc.). The model is a form of multi-variable linear regression. One goal is to use the results in the form of predicted uncontrolled NOx emissions to control the SNCR parameters, especially the amount of agent injected.
The model appears to do a reasonable job of following the NOx trends from the test facility. As such, it could have value as the basis for adjusting the SNCR reagent flow on the fly. My principal question is around the generality of the approach. It appears to be a bit of a brute-force curve fit, and as such would it work well for other facilities? Will a separate training exercise be required for each installation that is a different configuration? The presumed model is partly based on fundamentals (e.g., coal flow being linearly correlated with temperature). My one concern is that some of the processes are fundamentally very non-linear, especially Zeldovich NOx formation vs. temperature. It would be good to address the link between the model and the physics of NOx formation better.
Also, the performance of SNCR is critically linked to injecting the agent at exactly the right temperature, around 1250 K. Injection at temperatures outside this relatively narrow window results in reduced performance. If the injection temperature is significantly below the optimum, additional ammonia slip becomes a problem. Thus, although it is out of the scope of this work, the influence of the parameter on the temperature at the agent injection point will have an important influence on the performance of an integrated system.
The work appears to have been done carefully and correctly. The contribution is more the development of a semi-empirical tool and not the generation of new fundamental knowledge. The idea could have some practical utility, and that would be the point upon which the question of publication should lie. If the journal is comfortable with this, then the paper is publishable.
Some specific comments:
Line 91: NBC should be defined in both the text and the abstract as the text should be able to stand alone from the abstract. The authors follow this rule for SNCR, which is defined in both the text and the abstract.
Line 133: I’m not sure what is meant by “gas concentration”. Is this the total flow rate of gaseous combustion products? Please use the precise terminology.
Line 139: “Combustion effect of coal”. Again, this is not standard terminology. Is this the combustion stoichiometry?
Lines 151-156: Note that the fundamental effect of staged combustion on the reduction of NOx is the reaction of fuel nitrogen to N2, not the reaction of NO+CO. The evolution of fuel nitrogen from the coal under oxygen-starved conditions leads to N2 formation without NO as an intermediate. The efficiency quoted is, however, correct.
Text in the last three lines of p. 9 appears to be in need of repair?
Author Response
The authors are grateful for the helpful suggestions made by this reviewer.
In view of the questions raised by the reviewer, we added some explanations in the article as shown in line 370-374. For different production lines, the production parameters affecting the background concentration of NOx and their influence rules may be slightly different, which means that the conclusions obtained in this paper are not applicable to all production lines. In the popularization and application of the model, the above analysis process must be repeated to ensure the reliability of the results.
Influence of the parameter on the temperature at the agent injection point does have an important influence on the performance of an integrated system, and in the current SNCR system of cement production line, the temperature of reductant injection position is usually around 1173 K, and the research of this paper is based on this.
1) NBC in line 91 have been defined.
2) “gas concentration” in line 133 is the volume fraction of each component in the gas detected by the instrument.
3) “Combustion effect of coal” has been replaces as “Combustion efficiency of coal”, which is used to describe whether the pulverized coal entering the firing system is fully burned, for instance, whether there are products of inadequate combustion such as CO.
4) Thank you for the reviewer’s comments, lines 151-156 have been replaced as “Staged combustion is usually implemented at the bottom of the calciner, where the evolution of fuel nitrogen from the coal under oxygen-starved conditions leads to N2 formation without NO as an intermediate thus to reduce the NBC.”
5) Sorry for our negligence, text in the last three lines of p. 9 has been modified.
in [in] 介词在; 中; 于; 里; 其中; 里边; 在 ... 之内 定义 介词expressing the situation of something that is or appears to be enclosed or surrounded by something else. 形容词(of a person) present at one's home or office. 副词expressing movement with the result that someone or something becomes enclosed or surrounded by something else. 名词a position of influence with someone powerful or famous. 同义词 介词into; inside; into the interior of 形容词fashionable; in fashion; in vogue; voguish; stylish; in style; popular; (bang) up to date; up to the minute; modern; modish; trendsetting; chic; à la mode; de rigueur; trendy; all the rage; with it; cool; the in thing; hot; hip; happening; now; swinging; all the go 副词high; at its highest level; rising 示例 ask the lady in green
Round 2
Reviewer 1 Report
The revised version submitted by the authors include a few small changes that do not address the objections this reviewer posed on the previous round. Neither authors' reply refutes the criticism.
A linear regression model of a cement plant in a stable state (without the changes that those plants usually experiment), using data of just one and a half hour of operation, and validated with data from the following hour is a mere statistical exercise. And even in that favorable scenario, the variables of the model lose their physical meaning: A higher temperature in the kiln induces a higher motor current and more thermal NOx. The reduction that predicts the negative coefficient of x_9 is a mathematical artifact produced by the scarcity of data. (The decrease is the effect of more coal and incomplete combustion, what eventually compensates the increase of NOx due to higher kiln temperature). The conclusion is that the model is likely able to predict only the particular conditions of the experimental interval.
The authors say they have more experimental data. Fine, please let us know the values and include them in the regression to improve the model. They intend to do more testing in following paper, but the experiments must be done now to complete this paper. The problem is interesting (and complex) and the authors write well, still there are no shortcuts to the solution: You need more data, a better model and to demonstrate that it works under different conditions.
Author Response
The authors are grateful for the helpful suggestions and questions made by this reviewer.
The purpose of our writing is to put forward one technical route to achieve ultra-low emission of nitrogen oxides based on SNCR, and one of the keys to this process was the real-time acquisition of the NOx background concentration (NBC). One modeling method based on linear regression is proposed to realize the real-time acquisition of NBC in this paper, we are sure that there will be other better modeling methods to predict NBC more accurately under the various conditions and production lines. Ongoing research will be carried out to optimize the model, and the more results will be introduced in the following articles.
Question1: A higher temperature in the kiln induces a higher motor current and more thermal NOx. The reduction that predicts the negative coefficient of x_9 is a mathematical artifact produced by the scarcity of data.
Questions on the motor current of the kiln raised by the author was analyzed in the latest manuscript, as shown in line 424-439, for the 5000 t / D production line, NBC increases with the increase of the current of kiln motor x_9, it is consistent with theoretical experience and the reviewer’s point
The reason why the current of kiln motor of 12000 production line had a negative correlation with NBC is also analyzed based on the actual situation of the production line. Calorific value of the incoming coal of this 12000t / D production line is low and the flammability is poor due to the influence of coal supply. Excessive kiln coal is adopted to force calcination in kiln consciously thus to stabilize the quality of clinker. The insufficient combustion of excessive kiln coal ensures the high temperature required by the sintering zone in kiln, improves the calcination effect of clinker and increases the current of kiln motor. However, the unburned kiln coal will continue to enter the transition section of kiln with flue gas and burns under weak oxygen atmosphere, where the reductive CO gas can be generated, thus the thermal NOx generated in the kiln would be reduced and NBC value would be decrease. The above conclusion is contrary to the conventional knowledge, but it exists objectively in this 12000t / D production line.
Question2: They intend to do more testing in following paper, but the experiments must be done now to complete this paper. The problem is interesting (and complex) and the authors write well, still there are no shortcuts to the solution: You need more data, a better model and to demonstrate that it works under different conditions.
New data is added as a new separate Chapter 6 of the revised manuscript, which is about the soft measurement of NBC based on real data of one 5000t / D production line.
NBC prediction modeling was also carried out in one 5000t / D production line in Guangxi Province according to the same method and process as in Chapter 5, it is added as a new separate Chapter 6 of the revised manuscript.
The same conclusion as the 12000t / D model can be obtained that NBC of the 5000t / D production line was positively related with the temperature of outlet of calciner X3 and was negatively related with the temperature of smoke chamber X4. Models of the two production lines also shows some different conclusions, for instance, , NBC increases with the increase of the current of kiln motor X9 in the 5000t / D production line while the current of kiln motor X9 has a negative correlation with NBC in the 12000t / D production line. The analysis is carried out based on the process theory and the actual situation of the production line.
For the 5000-t/d production line, its NOx background concentration had positive correlations with the temperature of the calciner outlet, the current of the kiln motor and CO at preheater outlet and negative correlations with the temperature of the smoke chamber, pressure of smoke chamber and speed of high temperature fan. MAE between the predicted and measured values of the NBC was 14.8171 ppm, and the error between the maximum deviation value and the NBC was less than 5%.
Once again, we would like to express our heartfelt thanks to you and hope that this article can be affirmed by you.
Author Response File: 
Author Response.pdf
