Quantifying Multi-Parameter Dynamic Resilience for Complex Reservoir Systems Using Failure Simulations: Case Study of the Pirot Reservoir System

Round 1

Reviewer 1 Report

This subject addressed is within the scope of the journal. However, the manuscript in the present version contains several problems. Appropriate revisions should be undertaken in order to justify recommendation for publication.

1. 
   It is mentioned that systems analysis approach is used. What are the advantages of adopting this particular method over others in this case? How will this affect the results? More details should be furnished.
2.     For readers to quickly catch your contribution, it would be better to highlight major difficulties and challenges, and your original achievements to overcome them, in a clearer way in abstract and introduction.
3. There is a serious concern regarding the novelty of this work. What new has been proposed?
4. Abstract needs to modify and to be revised to be quantitative. You can absorb readers' consideration by having some numerical results in this section.
5. There are some occasional grammatical problems within the text. It may need the attention of someone fluent in English language to enhance the readability.
6. 
   The discussion section in the present form is relatively weak and should be strengthened with more details and justifications.
7. In conclusion section, limitations and recommendations of this research should be highlighted.
8. The authors have to add the state-of-the art references in the manuscripts.
9. It is mentioned that Pirot Water System, in the Republic of Serbia,is adopted as the case study. What are other feasible alternatives? What are the advantages of adopting this case study over others in this case? How will this affect the results? The authors should provide more details on this. 

Author Response

Dear Prof. dr. Jean-Luc Probst,

We wish to thank you and the reviewers for providing us with valuable feedback. We have considered in detail the comments by the reviewers and made changes to the manuscript accordingly. Please find below our responses (in italic) to the reviewer’s comments (in bold). The changes in the revised manuscript are highlighted in yellow.

We hope that you, the reviewers will find our manuscript suitable for publishing in the Water Journal after this revision.

Sincerely,

Milan Stojkovic

RESPONSE TO THE COMMENTS BY THE REVIEWER

1. It is mentioned that systems analysis approach is used. What are the advantages of adopting this particular method over others in this case? How will this affect the results? More details should be furnished.

Indeed, this point should have been elaborated in more detail in the original manuscript. Thus, now the authors have highlighted the advantages of adopting systems analysis approach used in the study (a revised paragraph between lines 72-80):

“This approach includes the development of the system dynamics simulation model with a causal loop that could be used to make decisions in water resources management and planning, testing the changes in input parameters which affect the operation of the multipurpose reservoir system. Moreover, it offers an opportunity for highlighting the role of using deterministic multi-model simulations which facilitates simulations taking into account feedback analysis of the effects of alternative system structures and control policies on system behavior [30]. The object-oriented system dynamics simulation approach allows for the incorporation of the high level of details of the complex reservoir systems. The system dynamics model facilitates investigation of nonlinear behavior in complex reservoir systems providing the values of variables at each simulation time step and in-sight how the changed water policies and operational rules affect the system as a whole [30].”  

2. For readers to quickly catch your contribution, it would be better to highlight major difficulties and challenges, and your original achievements to overcome them, in a clearer way in abstract and introduction.

We appreciate the comment. The authors revised the corresponding parts of the introduction and abstract. Major difficulties and challenges are elaborated in the revised paper with more detail. The introduced changes in the revised paper are given in the lines 11 - 14 (abstract):

“The objective of this research is to introduce a novel framework to quantify the risk of the reservoir system outside the design envelope taking into account the risks related to flood-protection and hydroenergy generation under unfavourable reservoir element conditions (system element failures) and hazardous situations within the environment (flood event).”

and 120 – 146 (introduction):

“The focus of this paper is two-fold; the first objective is to introduce a novel dynamic risk measure of the reservoir system – multi-parameter resilience linking the risks outside the design envelope of the confronting system drivers (risks related to flood-protection and hydroenergy generation); the second goal is to incorporate an explicit method within the system dynamic approach for modelling single and multi-component failures of the reservoir system elements.

Although the dynamic measures for risk assessment are already applied to estimate the risk of reservoir systems, the issue regarding the multi-parameter dynamic resilience of a complex reservoir system with an inner control loop and environmental system (e.g., river basin) is still not addressed. Therefore, the proposed approach helps provide the link between the complex multipurpose system and environment trading off among opposite system demands (flood-protection and hydroenergy generation) resulting in the single dynamic risk measure (multi-parameter resilience).

In addition, the behavior of the reservoir system elements over the mutual hazardous events (e.g. collapse of the dam, and/or collapse of any of its structural, mechanical or electric components, floods) alongside unfavorable conditions within the environmental system (e.g. floods) is not so far directly incorporated within the system dynamics approach. In this research, a novel object-based method introduces a concept of the functionality indicator defined for each reservoir system element to explicitly describe a physical drop of the element functionalities and the corresponding effects in the system operation.

 Therefore, a focus of this research is a proposition of a novel, universal approach for quantifying multi-parameter dynamic resilience for complex reservoir systems taking into account system element functionalities. Multi-parameter resilience brings an insight regarding hydropower and flood-defence system performance using the interconnected system variables and inner control loop, and providing the unique criterium for a reservoir operation strategy under multiple hazardous events like major flood, earthquake, leakage etc.”

3. There is a serious concern regarding the novelty of this work. What new has been proposed?

The authors responded to this comment within the response related to the previous comment. By introducing the changes within the introduction sections (lines from 120 to 146), the authors tried to emphasize the novelty of the presented study.

4. Abstract needs to modify and to be revised to be quantitative. You can absorb readers’ consideration by having some numerical results in this section.

Hmmm, an interesting point. The authors have included the numerical results in abstract. The introduced changes are given between lines 33-34:

“Discrepancy between the drop between multi-parameter resilience (from 0.851 to 0.935) and reliability (from 0.993 to 1) shows that static measure underestimates the risk on the water system.”

5. There are some occasional grammatical problems within the text. It may need the attention of someone fluent in English language to enhance the readability.

We sincerely regret for the English-related issues. The paper has been carefully edited. Hopefully, all the errors have been addressed.

6. The discussion section in the present form is relatively weak and should be strengthened with more details and justifications.

Highly appreciated comment. The authors revised the discussion section accordingly. Please find the improved discussion section between lines 629-681.

“From Figures 6 and 7, depicting water level variations and spillway flow rate as well as the changes in dynamic resilience, in the analyzed time period, it can be seen that the system dynamic model managed to adequately represent the effects of the single and multi-component failures. One of the key advantages of the novel explicit approach for the component failure modelling is, over stochastic ones [2], is the possibility of modelling the effects of the system investments. For example, if the system dynamics analysis based on the multi-parameter resilience assessment identifies a particular element (e.g. segment gates) as a weak point of the reservoir system, future in-vestment projects should consider the possibility of the revitalization of these gates or acquiring spare parts, which would reflect in the reduced failure duration. Now the same set of scenarios, with reduced failure duration for segment gates, can be run through the system dynamics model, to verify the effects of this particular investment decision.

The quantification and modelling of the dynamic resilience of a complex system are still under debate since the modelling steps are generally proposed for a particular system needed to mimic its nonlinear system behavior [34]. In general, the comparison of the results which quantify the dynamic resilience is difficult due to the following reasons [33]: various linked elements of a resilience assessment, and the selected simulation time step and total simulation period. However, the normalized form of a multi-parameter dynamic resilience can be compared with the static risk measures (e.g. reliability) obtained for the Pirot reservoir system, as the same models and failure scenarios are used for risk assessment.

Prior to the multi-parameter dynamic resilience assessment, the static risk measures are determined since they are commonly used to estimate water reservoir related risks. Considering the static measures provided in Table 5, similar results are obtained in literature [3], where the values of reliability are generally high (>0.945) compared to static resilience (0.0097-0.0275). Taking into account the changing climate conditions system resilience can also be lower than reliability in the assessment of the reservoir performance [32]. The relationship between static risk measures (Table 5) indicates the inverse proportionality between resilience and vulnerability. Resilience shows high values, while the system vulnerability seems to be low leading to the fact that the reliable reservoir systems are more resilient and less vulnerable [12]. There-fore, the failure scenario which preserves the high level of system performance is the one with high reliability or resilience values. In the terms of hydropower generation resilience, these scenarios are SC0, SC1 and SC2, while flood-protection resilience tar-gets scenarios SC0 and SC1. Other considered scenarios express a low level of static resilience, although the system reliability is still high.

By comparison with multi-parameter dynamic resilience, the static risk measures underestimate the risk stemming from the hazardous events, while dynamic resilience provides a successful tool that quantifies the risk over simulation time. For instance, the reliability of the system is in the range from 0.993 to 1, while the multi-parameter resilience ranges from 0.851 to 0.935 (Figure 8). The severe failure scenarios, for exam-ple, SC1 and SC3, slightly decrease the system reliability which is not in the line with the reduced system functionality which is indicated by dynamic resilience. However, the multi-parameter dynamic resilience does not reach the pre-disturbance level over-all simulation time steps even if the system reaches full performance level. This can be attributed to the fact that the disturbed reservoir system requires additional time to recover the system functionality [3].

Moreover, the multi-parameter dynamic resilience can assist to target the weak points of the Pirot reservoir system. The results in Figure 8 suggest that the overall functionality of the Pirot reservoir system is highly vulnerable to the increased leakage from the reservoir and failure of the spillway (SC2, SC3). Additionally, the failure of the spillway gate stresses the system functionality significantly (SC1). However, the system is still robust when the spillway gate at the compensation reservoir and Toplo-dolski tunnel fail (Figure 8).”

7. In conclusion section, limitations and recommendations of this research should be highlighted.

In the revised version of the manuscript, the authors described the limitation of the proposed approach alongside the recommendation for future work. The changes in the paper are introduced between lines 740 and 750.

“Future research can be oriented in two directions. First, to reduce the uncertainty related to functionalities of the critical system elements within the proposed concept, there-fore, the results from the numerical dam safety model needs to be incorporated within the concept for dynamic resilience quantification [35]. Such a numerical model is capable of providing more realistic behaviour of the dam under earthquakes, for instance, leaking from the reservoir and failure of the spillway gates may be determined on a physical basis rather than using a heuristic approach presented in this study. Second, the extraction of interpretable knowledge from a large amount of data gathered through the system dynamic model simulations under a number of generated failure scenarios can provide a solid basis for improvements in the dynamic resilience assessment. This can reduce un-certainties due to a limited number of failure scenarios considered.”

8. The authors have to add the state-of-the art references in the manuscripts.

The authors included the state-of-the art references in the revised paper (e.g. [32], [34], [35], [36], [37]).

9. It is mentioned that Pirot Water System, in the Republic of Serbia, is adopted as the case study. What are other feasible alternatives? What are the advantages of adopting this case study over others in this case? How will this affect the results? The authors should provide more details on this.

The authors have tried to justify in more detail the selection of the Pirot reservoir system. Please find the changes between lines 688-700.

“The proposed concept is applied at the Pirot reservoir system located in Serbia. This complex reservoir system is selected since it needs to satisfy several opposite demands, where the most important is flood-protection and hydroenergy generation. It lies at a flood-prone region in southeast Serbia where the snow-related processes have an impact on flood genesis. Moreover, it assists to cover “peak hours” when demand for electricity in Serbia are highest. Considering these issues, the proposed concept enables the assessment of multi-parameter dynamic resilience enveloping the aforementioned opposite demands and suggesting the adequate operation of reservoir systems under multi failures. Moreover, this concept is generic and offers application on alternative reservoir systems by changing the reservoir inflows, structure of the system elements and operational rules. It is worth mentioning that the application of the proposed concept to alternative study re-quires a redefinition of dynamic resilience measures following the system functions (e.g. water supply, drought management, environmental services and recreational activities).”

Author Response File: Author Response.docx

Reviewer 2 Report

The current manuscript deals with the resilience of water supply systems using failure simulations. The writing is OK and the reviewer had a few comments:

1) Formatting at lines 11-12 is strange (extra empty line?)

2) The abstract can stress the structure of the loops a little bit more. If possible, the novelty of the combination of models can be briefly highlighted in the abstract to attract more downloads and citations.

3) Line 39: Is it necessary to specify “multi-purpose water systems” here in the introduction. That appears to be over-limiting the scope of the paper. Also, the definition of “multi-purpose” is not clear either.

4) Lines 72-85: Please unify the font throughout the paper.

5) The introduction does not identify the knowledge gap in pertinent fields related to this study. Lines 105-118 do not have any citation to support the reasoning here.

6) The indentation of equations looks strange.

7) The part that describes the grand concept of the methodology is too short to show the novelty of this study. From the reviewer’s perspective, only lines 127-136 is dedicated to this purpose, and frankly, it does not provide much information.

8) Section 2.1 appears to cram several different things (weather generation, objective functions, etc.) in the same section. It is recommended to divide them into several smaller sections.

9) Hydrological modeling accuracy for calibration and validation tends to be low. For models calibrated by SCE, the R2 should be much higher.

10) The presentation of results is disjointed. Please generate a systemic way to present the results. Readers should be able to predict the structure of the “results” section based on how the “methodology” section is structured.  

Author Response

Dear Prof. dr. Jean-Luc Probst,

We wish to thank you and the reviewers for providing us with valuable feedback. We have considered in detail the comments by the reviewers and made changes to the manuscript accordingly. Please find below our responses (in italic) to the reviewer’s comments (in bold). The changes in the revised manuscript are highlighted in yellow.

We hope that you, the reviewers will find our manuscript suitable for publishing in the Water Journal after this revision.

Sincerely,

Milan Stojkovic

RESPONSE TO THE COMMENTS BY THE REVIEWER

1. Formatting at lines 11-12 is strange (extra empty line?)

The authors have resolved this issue.

2. The abstract can stress the structure of the loops a little bit more. If possible, the novelty of the combination of models can be briefly highlighted in the abstract to attract more downloads and citations.

We highly appreciate your valuable comment. The authors have revised abstract of the paper. Major difficulties and challenges are elaborated in order to highlight the novelty of study. The introduced changes in the revised paper are given in the lines 11 – 14:

“The objective of this research is to introduce a novel framework to quantify the risk of the reservoir system outside the design envelope taking into account the risks related to flood-protection and hydroenergy generation under unfavourable reservoir element conditions (system element failures) and hazardous situations within the environment (flood event).”

3. Line 39: Is it necessary to specify “multi-purpose water systems” here in the introduction. That appears to be over-limiting the scope of the paper. Also, the definition of “multi-purpose” is not clear either.

Thank you for your comment, as we realized an important issue. We are speaking about multipurpose reservoir system. Therefore, the authors changed the multipurpose water systems into multipurpose reservoir systems. Furthermore, we included a small explanation in the introduction section. Please find the explanation between lines 45-48.

“Multipurpose reservoirs have an important role in responding to natural disasters by controlling the runoff generated on a larger watershed scale [1]. The services provided by such reservoir systems are multifold providing water supply, flood and drought management, electricity generation, environmental services and recreational activities [38].”

4. Lines 72-85: Please unify the font throughout the paper.

The authors resolved this issue.

5. The introduction does not identify the knowledge gap in pertinent fields related to this study. Lines 105-118 do not have any citation to support the reasoning here.

Thank you for pointing out this issue. The authors revised the introduction section to address the gap while the knowledge gap in pertinent fields is previously described in introduction (lines 115-119). The introduced changed in the revised paper are given in the lines 120 – 146.

“The focus of this paper is two-fold; the first objective is to introduce a novel dynamic risk measure of the reservoir system – multi-parameter resilience linking the risks outside the design envelope of the confronting system drivers (risks related to flood-protection and hydroenergy generation); the second goal is to incorporate an explicit method within the system dynamic approach for modelling single and multi-component failures of the reservoir system elements.

Although the dynamic measures for risk assessment are already applied to estimate the risk of reservoir systems, the issue regarding the multi-parameter dynamic resilience of a complex reservoir system with an inner control loop and environmental system (e.g., river basin) is still not addressed. Therefore, the proposed approach helps provide the link between the complex multipurpose system and environment trading off among opposite system demands (flood-protection and hydroenergy generation) resulting in the single dynamic risk measure (multi-parameter resilience).

In addition, the behavior of the reservoir system elements over the mutual hazardous events (e.g. collapse of the dam, and/or collapse of any of its structural, mechanical or electric components, floods) alongside unfavorable conditions within the environmental system (e.g. floods) is not so far directly incorporated within the system dynamics approach. In this research, a novel object-based method introduces a concept of the functionality indicator defined for each reservoir system element to explicitly describe a physical drop of the element functionalities and the corresponding effects in the system operation.

 Therefore, a focus of this research is a proposition of a novel, universal approach for quantifying multi-parameter dynamic resilience for complex reservoir systems taking into account system element functionalities. Multi-parameter resilience brings an insight regarding hydropower and flood-defense system performance using the interconnected system variables and inner control loop, and providing the unique criterium for a reservoir operation strategy under multiple hazardous events like major flood, earthquake, leakage etc.”

6. The indentation of equations looks strange.

The authors have justified the equations.

7. The part that describes the grand concept of the methodology is too short to show the novelty of this study. From the reviewer’s perspective, only lines 127-136 is dedicated to this purpose, and frankly, it does not provide much information.

We understand the concern. The authors have tried to highlight the novelty of this study in the Introduction section. The corresponding changes in the manuscript are given in the response related to the comment no. 5. Also, authors shortly described the novelty of the applied concept in the section of methodology (lines 157-160).

8. Section 2.1 appears to cram several different things (weather generation, objective functions, etc.) in the same section. It is recommended to divide them into several smaller sections.

Thanks. The authors divided the text into several sections (e.g. climate data generator, hydrological modelling, hydrological model calibration) following the reviver’s suggestion.

9. Hydrological modeling accuracy for calibration and validation tends to be low. For models calibrated by SCE, the R2 should be much higher.

The authors agree with reviewer that the agreement between the observed and model flows are relatively low. Therefore, they try, in the revised paper, to discuss the reason led to lower agreement. Please find the justification between lines 426-435.

“Results of hydrological model calibration and verification for selected hydrological stations (h.s. Visočka Ržana – Dojkinačka river, h.s. Pakleštica – Visočica river, h.s. Pirot – Nišava river) are depicted in Figure 4. Modelled daily flows show better agreement in the calibration phases with the observed values, while the agreement in the verification phase is somewhat lower. It is worst noting that hydrological modelling is performed using available daily precipitation sums over the Nišava river basin in Serbia, while the most upstream parts of the river basin (subbasins 14, 15, 16, 33) are not covered by the precipi-tation monitoring network (Figure 3). A lack of measured data results in a somehow lower agreement between observed and simulated daily flows at the sites of hydrological stations used.”

10. The presentation of results is disjointed. Please generate a systemic way to present the results. Readers should be able to predict the structure of the “results” section based on how the “methodology” section is structured.

Thank you for the comment. The authors rearranged the result section in the same manner as the methodology section. Moreover, several subtitles are introduced as for the introduction section following the reviewer’s previous comments. To improve presentation of the results, the authors also introduced the discussion section. Please find the introduced section between lines 629-681.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Authors revised properly.should accept now.

Reviewer 2 Report

Thanks for providing an updated manuscript. It looks good. The only thing is that some typos are still present, such as "worst" at line 430. Please proofread the whole manuscript again.