Coupling a Chlor-Alkali Membrane Electrolyzer Cell to a Wind Energy Source: Dynamic Modeling and Simulations

Round 1

Reviewer 1 Report

Title: Coupling a Chlor-Alkali Membrane Electrolyzer Cell to a Wind Energy Source: Dynamic Modelling and Simulations

Article Type: Full length article

Manuscript Number: Energies-1491073-v1

This paper describes a dynamic model of a chlor-alkali membrane cell was developed to assess the flexible potential of the membrane cell. The dynamic modelling and simulation results show that the CA plant can indeed be flexibly operated in the future energy system.

My recommendation is that the authors carefully consider the below points, revise appropriately

 

1. The authors should consider some representative word in the keywords.
2. Page 1, line 43~45; “…. the grid to remain at 50 Hz” The authors may consider provide us the general voltage as well as frequency in Germany for reader from different country.
3. Page 4, line 141~142; my suggestion is that the authors may give us some explanation about the effect of side product Cl₂ in anode.
4. Page 21, line 547~548; my suggestion is that the authors may give us deviation data to support the result in “…. in good agreement with experimental data and published model values……”

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Summary:

The presented paper describes a dynamic model of a color-alkali electrolysis cell which is capable of describing the concentrations of water/NaCl and water/NaOH in the regrading cell compartments, the produced chlorine and hydrogen flows, the unwanted evolution of oxygen and the cell voltage. The electrochemical model is compared with published steady-state experimental data. Finally, two case studys are performed where the dynamic behavior of the model is investigated.

Even though the article is well made in large parts, serious flaws are present. This is especially related to the claimed improvements over previously published models and the real world applicability of the model.

Major comments:

- The review part needs to be revised:

1) Some relevant recent publications in the field of dynamic chlor-alkali electrolysis models are not considered here (e.g. Weigert et al. (2021)). 
2) In line 117 and following the authors mention the improvements over earlier publications, however most of the claimed improvements are already included in e.g. Otashu et al. (2019) or Weigert et al. (2021). The only novelty in the presented article is the modeling of the oxygen evolution.  Nevertheless, an energy balance is missing compared to the earlier publications, which is also not mentioned. This needs to be clarified.
3) In line 108, the authors mention that in Budiarto et al. only the electrochemical model is validated with steady-state data. However, the same is true for the validation of the presented model. This needs to be clarified.
4) In line 125 the authors claim that the complete model is validated in the presented work. However in section 3.1. only the electrochemical model (cell voltage, membrane potential) is validated.

- The modeling part needs to be revised:

1) In line 114 the authors say, that the temperature constraints limit the load variations, but no energy balance is included in the model. Please clarify why this is the case.
2) Due to the model structure, the faradaic efficiency and the membrane permselectivity have to be equal in the model. This can be assumed as long as no acidification in the anolyte compartment is used to prevent oxygen formation. However, since oxygen formation has to be prevented in industrial CA processes acidification is widely used. Please explain why a model structure like this is valid for describing a "state-of-the-art CA plant".
3) The presented model includes dynamic component balances, dynamic mass balances and differential equations for the concentrations. This is very confusing since the different variables can be converted into each other just by using the molar masses and densities. It would be sufficient and would increase the clarity of the model if only one class of mass/component/concentration balances would be presented together with correlations for the densities.
4) Please go more into detail regarding the control of the system. The mentioned water feed rate (line 384) is not defined in the model. What is the concentration of NaOH in the incoming diluted NaOH stream before the dosing of the mentioned water feed? Is there a NaOH recycle considered in the model?
5) Please explain how the model is solved. Are all equations solved simultaneously or are the two model parts solved sequentially as shown in figure 3?

- The results and discussion part needs to be revised:

1) The deviations between the presented model and the published experimental data need to explained in more detail. What do the authors think, causes the deviations? Is there a way to reduce the deviations?
2) If the simulation approach in sequentially, how do you solve the system if the power profile is used as inputs (section 3.2.)?
3) In line 501 and figure 11 the authors use a current density ramp rate of 0.1 as the bounds for the normal operation of the CA plant. Please go more into detail where this number comes from. In the reviewers opinion this is not stated in O'Brien et al. (2007).
4) In section 3.2. the authors say that the electrolyte concentrations need to be maintained at a constant value to increase the lifetime of the membrane. This is also true for the cell temperature. Please explain why this is not considered in the case study.
5) In section 3.2. the authors claim that the concentrations can be effectively controlled and thus a CA plant can be flexible operated. However, are the resulting profiles of the manipulated variables realistic? Could a profile like this be applied to a real CA plant?

Minor comments:

Line 75: Is a whole paragraph on outdated diaphragm CA necessary?
Line 262: Is this really the faradaic efficiency instead of the permselectivity?
Line 278: The vapor pressure of water (equation 41) is already defined in equation 17
Line 300: In equation 50 it should be NaOH instead of NaCl in the subscripts.
Line 384: What does this mean: "... is adjusted by the current density to control the exit caustic concentration."
Line 420: What does this mean: "different solutions on both sides of the membrane."

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Summary

The authors provide a dynamic model of the chloralkali electrolysis. The chloralkali is of significant interest for demand response and has been looked at by several research groups in this context. The content of the manuscript would therefore certainly be within the scope of Energies.

The authors claim that their model includes some major improvements compared to other models published between 2017 and 2019. In the manuscript, they discuss their model extensively and provide steady-state and dynamic simulation studies to analyze its behavior. Finally, they provide an interesting, but questionable (in terms of a real application of demand response) case study in which they use the model for dynamic simulations to determine whether the electrolyzer could operate on the  power load changes of a wind turbine.

However, most of the major improvements the authors announce have already been included in other recent publications. Given that many of these publications have been released in 2021 and the newest references in the manuscript are from 2020, one has the impression that the authors were unaware of them. Unfortunately, this significantly diminishes the significance of the manuscript's content.

Scope

• The content of the manuscript is relevant for the readers of Energies
• In the reviewer’s opinion, the current manuscript does not go significantly beyond the scope of other recently published contributions.

Language-related aspects

• Language-wise, there are a few minor issues that are pointed out below but overall, the manuscript is well written
• Line 36: “operating” instead of operation
• Line 38: removed produced
• Line 61: understanding of
• Line 96: “formulated” instead of “made”
• Line 97: “are” instead of “being”
• Line 134: “in the section above”
• Line 135: “by using a … (CEM)”
• Line 162: “formulated” instead of “made”
• Line 171: space between “gas” and bracket
• Line 194: comma instead of period
• Line 219: space between Equation and (18)
• Line 246: space between “gas” and bracket
• Line 251: “cathode” instead of “anode”
• Line 404: “The model was solved”
• Line 549: “seem” or “are”?

Introduction

• Line 29: Why do the authors cite a goal from the past?
• Line 35: Do the authors talk about DSM or demand response?
• Line 41-44: What type of control reserve power do the authors talk about? I assume they refer to the German balancing market. There are still several options for balancing power. Some of them are not necessarily faster than, say, the day-ahead market.
• Line 54: [3,13] instead of [3],[13]
• Line 58-59: What are these specifically? Will they be of actual relevance or are they so far from actual operating points that they do not play a major role?
• Line 67-88: If the technology is outdated, the authors could simply state that with a respective reference and remove these two paragraphs as their content is not relevant for the CA technology.
• Line 100-114: The literature review seems to be lacking a significant share of the published literature on the topic: For example, Weigert et al. (2021) (10.1016/j.compchemeng.2021.107287) recently published a detailed dynamic model and compared it with industrial plant data. The model was also already used for optimal control of the CA electrolysis in the context of DR (10.1021/acs.iecr.1c01360). Furthermore, the groups of Baldea and Mitsos published several articles on the CA electrolysis, the latter focusing on the use of oxygen depolarized cathodes. Other contributions were published by Richstein and Hosseinioun (10.1016/j.apenergy.2020.115431) or Baetens et al. (10.3390/en13215675). The authors may argue that not all these contributions deal with rigorous models, but given that they look at demand response in general, it seems like they should put some more effort into a thorough literature search. Looking at the work published by Otashu et al. (2018) and Weigert et al. (2021), the authors should outline the significant improvements with respect to the other publications in more detail. For example, brine and caustic concentrations, migration through the membrane, OH- transport from cathode to anode, and the evaporation of water are already part of the model by Weigert et al. (2021). Both Otashu et al. and Weigert et al. additionally formulated dynamic energy balances, which are missing in the current manuscript.
• Line 125: Why did the authors choose to model the CA electrolysis under isothermal conditions? The impact of temperature drop or increase might be significant in the case of fluctuating current density. In addition, as the authors states themselves, Otashu et al. identified temperature constraints as significant limitations for load variation.
• Line 125-126: The authors criticized (rightfully) that Budiarto et al. only used steady-state data for model validation, but then the authors do the same in their manuscript. Could you please elaborate?
• Throughout the introduction, the authors do not clearly state the specific novelties in the manuscript.
• Could the authors please provide a short description of the structure of the manuscript at the end of the introduction?

Main part

• general comment: Why do the authors use upward symbols for variables? Usually, variables are set in italics.
• general comment: why did the authors use * instead of a dot for multiplication?

• general comment: shouldn’t Section 2.3 be Section 2.2.3? Controllers are part of the model, aren’t they?
• general comment: I would recommend naming Sections 3.1 and 3.2 Steady-state Simulations and Dynamic Simulations. The modeling part is Section 2.

• Line 138-146: How were the inlet and outlet concentrations of brine and soda selected? Did they take the values from other publications?
• Fig. 1: Where did the authors take the flowsheet from? Is their control strategy – using fresh water for temperature control – typical of these plants? Could the authors please briefly explain that?
• Line 151: Could the authors please provide a reference for this statement (inefficiencies due to OH- migration).
• Line 153: Is this side reaction part of the model equations? If yes, where is it?
• Line 157: The authors state that their model is an ODE system. However, there are many algebraic equations below, which indicate that it is in fact a DAE system. Could the authors clarify?
• Fig. 2: Fig. 1 and Fig. 2 basically show the same process. The authors are therefore encouraged to combine both figures.
• Eq. 6: The authors should add where the factor 0.5 come from.
• Line 177: “decomposition of water” is irritatingly formulated given that water is formed in Eq. 4. The authors should reformulate.
• Line 203: Based on which data or references assume the authors a value of 96 %?
• Line 205: Please do not use lumped citations.
• Eq. 13: The authors should explain the factor 4
• Line 208: The authors should rephrase the statement as an assumption
• Line 209-211: This argument is irritating: At the beginning of the manuscript, the authors claimed the consideration of the formation of oxygen to be a significant improvement compared to other models. Now they basically neglect the oxygen’s influence completely. Please explain this.
• Eq. 213: This equation seems to be wrong. The partial pressure is, by definition, y_i * P. The equation would thus read y_H2O * P = y_H2O / y_Cl2 * y_Cl2 * P, i.e., the equation is superfluous. The same applies to the respective terms in the cathode compartment.
• Line 217: Again, please use no lumped citations. You certainly do not need 3 references for the Antoine equations.
• Eq. 17: Please do not put the parameters in the equations, but compile a table in which the parameters are listed. Instead use parameters in the equations. Also add these parameters to the list of symbols.
• Eq. 18: MacMullin pointed out that this equation is only applicable if y > 3. Did the authors verify this in their simulation studies?
• Eq. 19: There are several questions. First of all, molality is mole solute per mass of solvent. Secondly, where does the factor 1000 come from? I assume it is because Eq. 18 yields R in the unit mol per kg? The authors should clarify that.
• Eq. 21 + 22: Why are both mole and mass balances formulated? If mole balances are known, they respective mass flows could directly be computed. The same question applies to Eqs. 25, 45, 46, and 49.
• Eq. 27: Why do the authors use R_NaCl again in this equation? The variable is already used.
• Eq. 51: Why is R_NaOH used again here? The variable has already been used.
• Line 309-310: Based on which reference make the authors this statement regarding the share of galvanostatic operated plants?
• Eqs. 59, 60, 61: The equations cannot be correct because the arguments of the logarithms are not dimensionless. Is this also incorrect in the model?
• Eqs. 62-67: The authors are encouraged to move such equations to a supplementary material or completely remove them as they can be found in the original reference.
• Eqs. 62 and 68: Again, did the authors verify that the molality is always within the stated bounds?
• Line 338: lumped citations.
• Line 337-343: There is no point in putting the numbers in the text and then also include them in Table 1. Please use the table and add the references there.
• Line 358-359: Could the authors please add a justification why they neglect E_eled?
• Line 375: The original source of Eq. 72 is not ref. 11. Could the authors please cite the original reference?
• Table 2: Could the authors please add the respective setpoints to both controllers? How were the controller parameters determined?
• Line 383-385: If the anolyte concentration is controlled using the inlet flow, could the cell run dry? Do the authors see any disadvantages with this control scheme?
• Table 3: What is the purpose of Table 1 if the parameters are repeated in Table 3, anyway? Please combine both tables.
  
  The reviewer would also like to point out that the given ramping rate is used as an example in O’Brien (2007) and should not the be seen as a maximum rate as seems to be indicated by the authors.
• Line 390-393: Please rephrase the sentences. They seem repetitive.
• Figure 3: Could the authors explain how exactly the outlet flows are calculated? The fact that they are algebraic variables would indicate a higher-index problem. On the other hand, the authors seem to assume that the liquid volume in the cells remains constant. This is very unclear in the model description.
  
  In addition, the figure uses the word “estimate”. It is unclear why this term is used. There should be an iterative loop, which ensures that all equations are solved accurately.
• Fig 4a: Could the authors please provide results for the experimental results by Berger et al. (same composition, etc.) and Takahashi. Otherwise, it is not really possible to compare model and experimental data
• Fig 5: The axis labels for the units are inconsistent. The authors are also encouraged to think about their scheme. The colors and color differences will probably be hard to read and differentiate.
• Line 435-437: Does this mean that the controllers were turned off? Otherwise, the inlet flows would have to change.
• Fig. 8 and 9: While these results are certainly interesting, the authors still assume a constant temperature. How would the results change with a changing temperature? Would this effect still be relevant?
• Section 3.2: This is, in the reviewer’s opinion, a problematic section. The authors use the profile of a single wind turbine to demonstrate the application of the CA electrolysis. However, how realistic is such a scenario? Usually, the wind turbine would feed electricity into the power grid, which represents a mix of various electricity sources. The grid operator would then balance the net frequency based on the overall available power and not based on a single wind turbine. If balancing power is required, the operator of the electrolysis would have some response time until the new load must be achieved. How would the authors respond? The authors are encouraged to rethink this case study and potentially adapt it to the conditions that apply to DSM in reality.
• Fig. 10: Could the authors also include the composition of the liquid effluent from anolyte and catholyte here instead of in Fig. 12. Power, current density, and voltage are not that informative as they basically coincide.
  
  Also: The authors state a possible range for the current density in Table 3. In Fig. 10, this range is clearly violated.
• Line 501: The reviewer would like to stress again that for a reasonable demand response scenario the power ramps of one particular plant are probably not useful. The ramps are rather the result of a particular balancing market and the time that is allowed to activate the balance reserve.

Conclusion and outlook

• Line 539-546: In the reviewer’s view, none of the improvements of the model compared to the work by Otashu et al. or others have been clearly highlighted, nor has their impact on the results been demonstrated or discussed. The authors also do not provide a reasoning why they include minor additions, such as the formation of oxygen at the anode, but completely neglect a dynamic energy balance.

Formal aspects

• the authors should put all symbols in alphabetical order to make it easy to find a specific symbol
• the authors should differentiate between abbreviations, Latin symbols, Greek symbols, subscripts, etc. in their list of symbols
• unit of molar flow rate (mol per L) should be mol per second
• chemical formulas, i.e., Cl-, Cl2, H2, H+, OH-, O2, NaCl, NaOH, and Na+ are self-explanatory as chemical compounds and should therefore be removed from the list of symbols
• B-V, IEC, and MEA seem to be unused abbreviations and should therefore be removed
• wt% is also not required to be part of the list of symbols
• Second in unit of Faraday’s constant should be small and not capitalized
• R_NaCl, U_NaCl and Z_NaCl are missing
• R_NaOH, U_NaOH, J_NaOH, and Z_NaOH are missing
• molalities should have the unit mol per kg
• instead of a_an,NaCl and a_cat,NaOH, just write a or a_k for activity: The other subscripts are already explained. Moreover, the activity of water is missing otherwise.
• Space missing in unit of i_0
• R_elec, i.e., the resistance is missing
• Small L (length of resistance) is missing
• The symbols alpha and delta seem to be unused

Text

• Section 3.2 exists twice

Figures and Tables

• Figures are in general of sufficient quality although the authors could think about the type of the used colors
• Sometimes, figures in black and white with different line types could suffice

References

• Reference 3 is missing title, volume number, page number, and DOI
• Reference 4 contains errors in the title
• Reference 7 is missing page numbers
• Reference 14 should be Euro Chlor, also a link with access date should be included
• Reference 16 is missing volume number, page number. Please also use the DOI of the original work
• Reference 25 is missing remaining authors. In addition, it is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 26 is missing the remaining authors.
• Reference 29: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 38: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 39: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 42: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 44 cannot be found under the given link. Also, bibliographic information is incomplete
• Reference 50: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 60: It is unclear what type of document it is – article, book, etc. Please add the necessary information
• Reference 61 seems to be missing page numbers

Appendices

• The authors are encouraged to consider outsourcing some of their less important equations into an appendix or supplementary material.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Summary

The reviewer would like to thank the authors for their improvements regarding the manuscript. Most of the remaining item are minor things. However, the structure and scenario of the final case study is still not suitable. More detailed comments on this issue are given below. The authors should consider to bound the current density between reasonable values. Another issue is that DR applications would also require a load shift to average the product. This would also be relevant in the context of this case study.

Language-related aspects

• Line 409: "resistance" instead of "resistivity"

Introduction

• Line 36: If the authors state they talk about DR, why are are they also referring to DSM? They should make clear whether these are synonyms in their manuscript or - if not - remove DSM.
• Line 41-44: These lines are still unnecessarily restrictive. First of all, DSM includes much more than DR. Secondly, control reserve is not necessarily the fastest form. Buying or selling electrical power on markets can also be very fast and is included in DR as well. The authors should rephrase this just to make it clear.
• Line 60-64: Here is now a major contradiction in the manuscript: The authors point out that too low current density will result in damage or will increase the chlorine concentration in caustic. On the other hand, the authors removed the operating range of the current density in Tab. 3 so that their case study does not need modification. It cannot be both. The authors should modify their case study accordingly.
• Line 72: The authors kind of contradict themselves here regarding the number of studies given the quite large number of references they cite afterwards. They could emphasize more their very mechanistic approach, which seems to be more detailed compared to other publications.
• Line 80-94: It is still not completely clear why the authors want to keep this section. If indeed the fundamentals from these studies are so relevant, the authors should be more specific which parts are relevant for their proposed model.
• Line 125-135: The authors should give state convincing disadvantages of the models by Otashu and Weigert. Otherwise, it is unclear why this manuscript should be published.
• Line 125: Why did the authors choose to model the CA electrolysis under isothermal conditions? The impact of temperature drop or increase might be significant in the case of fluctuating current density. In addition, as the authors states themselves, Otashu et al. identified temperature constraints as significant limitations for load variation.
• Line 146: Not a sentence.
• Line 136-147: Again, the authors contradict themselves here. In this paragraph, they state that they assume temperature control to be not an issue. This is in disagreement with the findings of both Otashu and Weigert. On the other hand, they want to study temperature in future work because it might have a significant impact. It cannot be both. The authors should rethink their argumentation.

Main part

• Fig. 1: The authors responded that they added an explanation of the flowsheet and their control scheme. However, I was unable to find it in Section 2.3.
• Eq. 20 ist the combination of Eqs. 14 and 15 and can therefore be removed.> 3. Did the authors verify this in their simulation studies?
• Eq. 21 + 22: I would again recommend to not use both dynamic mole and mass balances. This can cause numerical issues in the event that mass and mole balance are not consistently initialized. In addition, this approach only works under the implicitly assumed constant cell volume. Using an algebraic mass balance in the model would be much more intuitive.
• Eqs. 62-67: I would again recommend moving these equations into an appendix. Especially, since the temperature is fixed anyway and the given parameters are therefore constants. Readers will not have a better reading experience only because these equations exist.
• Your response 70: You said that the cell volume remains constant because you assume density to remain constant. However, these two are non-causal. The volume could still change if the holdups change. Therefore, you must specify the constant cell volume as an assumption.
• Response 73: Thank you for the clarification. However, I do not see the changes in the manuscript.
• Response 74: Again, it is not possible to assume the good control of plants and then state that the dynamic energy balance is important and should be added. It cannot be both.
• Response 75: The parameters of this study do still not really make sense. The authors outline in their introduction that there are limits regarding dynamic operation. In the last case study, they completely neglect this and allow the current density to vary between zero and maximum power. Because this case study is so far from reality, the authors should reconsider their case study (e.g., by using at least a suitable load range) or potentially remove it from the manuscript and instead focus on their other results.
• Response 76: Simply removing the existing operating limits from the manuscript is not a good solution in this case. The operating will remain. Given that the authors want to show that the CA electrolysis may operate under realistic conditions, this case study should be improved or removed to obtain reasonable results.

Author Response

We would like to thank the reviewers for this second reound of review. We appreciate the comments and suggestions to improve the quality of our manuscript. We have done our best to provide a point-by-point response to all the comments which are marked in yellow in the new revised version of the manuscript.

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 3 Report

No additional comments