An Operation Scheme Generation Method for Nuclear Power Plant Operation under the Condition of No Operating Procedures Guided
Round 1
Reviewer 1 Report
The paper presents a technique to identify success paths for achieving system objectives, which is potentially useful in unexpected situations when no operating procedures is prepared in advance. I find the manuscript falls a bit short in several areas, upon reflection and implementing changes, I believe will improve the quality further.
1. On literature review:
While the authors refer to several report to indicate how the risk of human error is significant in NPP operation and how the chance of human error may increase in unfamiliar environments, it seems that a review of works related to the proposed methods has been missed. The most relevant concept is called Synthesis of Operating Procedures. Authors may find a lot of action searching methods both in normal and emergency operations in journals such as Computers & Chemical Engineering. In particular for MFM application, there is a research line of using causal rules between functional primitives for backward searching of counteraction sequences (compare to what author proposed forward searching), which can trace back to [Gofuku and Tanaka (1999), Application of a Derivation Technique of Possible Counter Actions to an Oil Refinery Plant], until a recent work that uses a set of new types of rules in MFM [Song et al (2020), Model-based and rule-based synthesis of operating procedures for planning severe accident management strategies]. Regarding path searching in single function structure, author may refer to [Song et al (2019), Synthesis of Valve and Pump Operations in Complex Plants by Using Functional Modeling].
2. On methodology:
· The core method is essentially a forward searching process starting from source function. Compare to the backward searching which starts from a given objective, the biggest problem is how to ensure that a path starting from a particular source function can lead to the objective achievement. Therefore, authors need to explain their rationality for such consideration.
· Figure 4 does not comply with MFM syntax, which is not a serious issue though. In general, a flow structure can only have one main function, i.e. the function directly contributing to achieving objective. When there are more than one function related to objective, e.g. tra3 and tra4, authors may consider change of operating modes or a higher-level function that can link to both tra3 and tra4, then connect it to objective.
· In case study, authors also consider to analyze potential hazardous consequences of identified success path. Do you think this analysis can be integrated into the method with MFM?
· Also in case study, it seems that authors use some domain knowledge to re-construct paths to form a new path that can be actually implemented. This is reasonable because different identified paths can represent different aspects of goal achievement, energy or mass, which need to be accomplished in a special order. The question is whether there can be a systematic or formal way to do this ‘re-construction’.
· Can the whole framework be automated by algorithm?
3. On text:
· In Abstract, MFM models system through a means-end analysis. However, the path identification is not accomplished through a means-end analysis. In fact, it is more of a searching approach.
· In Page 2 – Line 66, what do you mean by ‘prevent human error from human error’?
· In Page 4, Figure 3, reference?
· In Page 8, please check the tags of relations connecting objectives. It seems that they do not match the corresponding functions. This also goes back to the question on methodology, which function is exactly related to objective?
Author Response
Point 1: While the authors refer to several report to indicate how the risk of human error is significant in NPP operation and how the chance of human error may increase in unfamiliar environments, it seems that a review of works related to the proposed methods has been missed. The most relevant concept is called Synthesis of Operating Procedures. Authors may find a lot of action searching methods both in normal and emergency operations in journals such as Computers & Chemical Engineering. In particular for MFM application, there is a research line of using causal rules between functional primitives for backward searching of counteraction sequences (compare to what author proposed forward searching), which can trace back to [Gofuku and Tanaka (1999), Application of a Derivation Technique of Possible Counter Actions to an Oil Refinery Plant], until a recent work that uses a set of new types of rules in MFM [Song et al (2020), Model-based and rule-based synthesis of operating procedures for planning severe accident management strategies]. Regarding path searching in single function structure, author may refer to [Song et al (2019), Synthesis of Valve and Pump Operations in Complex Plants by Using Functional Modeling].
Response 1: Thank you very much for your comment. I have added the review of works related to the proposed methods in the Success path identification method(see Page 4, Lines 149-159).
Point 2: The core method is essentially a forward searching process starting from source function. Compare to the backward searching which starts from a given objective, the biggest problem is how to ensure that a path starting from a particular source function can lead to the objective achievement. Therefore, authors need to explain their rationality for such consideration.
Response 2: Thank you very much for your comment. The reason of why this paper utilizing a forward searching from source function is that we don’t consider all success paths have been potentially connected to the objectives during MFM modeling. Instead, we assumed that some paths that was considered being impossible to achieve the objectives in the design phase can be utilized during NPP operation. The revised paper explained the reasons in the discussion part (see Page 14, lines 328-344).
Point 3: Figure 4 does not comply with MFM syntax, which is not a serious issue though. In general, a flow structure can only have one main function, i.e. the function directly contributing to achieving objective. When there are more than one function related to objective, e.g. tra3 and tra4, authors may consider change of operating modes or a higher-level function that can link to both tra3 and tra4, then connect it to objective.
Response 3: Thank you very much for your comment. I have modified Figure 4. (see Page 5, Line 181).
Point 4: In case study, authors also consider to analyze potential hazardous consequences of identified success path. Do you think this analysis can be integrated into the method with MFM?
Response 4: Thank you very much for your comment. The proposed method might be used for a potential path (to either success or failure) analysis method in the future, which may supplement the path identification scope of backward searching.
Point 5: Also in case study, it seems that authors use some domain knowledge to re-construct paths to form a new path that can be actually implemented. This is reasonable because different identified paths can represent different aspects of goal achievement, energy or mass, which need to be accomplished in a special order. The question is whether there can be a systematic or formal way to do this ‘re-construction’.
Response 5: Thank you very much for your comment. The authors don’t think the proposed method can be a systematic or formal way to re-construct MFM models. Field knowledge is needed to supplement by plant personnel. (see Page 14, Lines 344-349).
Point 6: Can the whole framework be automated by algorithm?
Response 6: Thank you very much for your comment. No, at the moment it is a semi-automated framework. Field knowledge is required to form complete success paths. (see Page 14, Lines 349-352)
Point 7: In Abstract, MFM models system through a means-end analysis. However, the path identification is not accomplished through a means-end analysis. In fact, it is more of a searching approach.
Response 7: Thank you very much for your comment. We have improved the abstract (see Page 1, Lines 19-20).
Point 8: In Page 2 – Line 66, what do you mean by ‘prevent human error from human error’?
Response 8: Thank you very much for your comment. We have modified it to “human error”(see Page 2, Line 71).
Point 9: In Page 4, Figure 3, reference?
Response 9: Thank you very much for your comment. We have repainted Figure 3. (see Page 5 Line 160 )
Point 10: In Page 8, please check the tags of relations connecting objectives. It seems that they do not match the corresponding functions. This also goes back to the question on methodology, which function is exactly related to objective?
Response 10: Thank you very much for your comment. The model in the figure was modified. (see Page 8, figure 7, Line 249)
Author Response File: 
Author Response.pdf
Reviewer 2 Report
1) Please improve the text quality and try to avoid some mistakes (e.g. line 40 - to "reduce", line 45 conditions "are", etc. )
2) The quality of Figure 3 is low, try to convert it to vector format.
3) The mathematical description of the proposed MFM model is missing.
4) Please improve the conclusion and abstract part to avoid misleading sentences.
5) It could be interesting to propose conceptual function that can describe the risk dynamics that involves the variables of the unsafe human acts.
6) Please indicate how the proposed approach correlatives with the safety standards by the IAEA.
7) The paper is well structured and contains scientific novelty.
Author Response
Point 1: Please improve the text quality and try to avoid some mistakes (e.g. line 40 - to "reduce", line 45 conditions "are", etc. )
Response 1: Thank you very much for your comment. We have made some revisions to the text and hope that this time it meets your requirements. (see Page 2 Lines 44,49 )
Point 2: The quality of Figure 3 is low, try to convert it to vector format
Response 2: Thank you very much for your comment. We have converted Figure 3 to a vector format to improve its quality. (see Page 5 Lines 160 )
Point 3: The mathematical description of the proposed MFM model is missing
Response 3: Thank you very much for your comment. We have add the mathematical description of the proposed MFM model. (see Page 5 Lines 162-173)
Point 4: Please improve the conclusion and abstract part to avoid misleading sentences.
Response 4: Thank you very much for your comment. We have improved the conclusion and abstract.(see Page 1 Lines 19-20, and Page 14 Lines 328-352)
Point 5: It could be interesting to propose conceptual function that can describe the risk dynamics that involves the variables of the unsafe human acts.
Response 5: Thank you very much for your comment. We will consider your valuable comment in our future work.
Point 6: Please indicate how the proposed approach correlatives with the safety standards by the IAEA.
Response 6: Thank you very much for your comment. We described the relationship between the method and the safety standards of the IAEA in the discussion (see Page 14, Lines 314-316 ).
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have properly answered the questions from reviewer. I think the manuscript can be accepted in the current version.
