Solving the Integrated Multi-Port Stowage Planning and Container Relocation Problems with a Genetic Algorithm and Simulation

Round 1

Reviewer 1 Report

This paper presents an integrated methodology for the joint resolution of two important tasks in port yard, namely the Container Relocation Problem (CRP) at yards and the Multi-Port Stowage Planning Problem (MPSP) for the charge and discharge of ships. More in detail, the proposed approach combines simulation, a genetic algorithm, and a rule-based heuristics, with the final aim of minimizing the total number of relocations for both the yard and the ship. The method is tested on several instances of growing complexity. Overall, results are sound and promising, as well as future research directions are clearly highlighted at the end of the paper.

The paper is interesting, well written and well-structured enough. The aim of the contribution is clear, and the manuscript has a practical relevance.

The introduction is concise, but clear and effective. Although the reference list is complete and updated enough, I think the literature review and the state of the art description could be improved. In addition, authors should better relate their work to other researches, emphasizing the paper contribution and its positioning within the existing literature, and highlighting its novelty and main advantages. In particular, I do not agree with the statement of the authors that in the relevant literature “the decisions of different interests of the diverse agents have not been incorporated”. I suggest the authors to also recall some papers dealing with the same operations when performed in multimodal and intermodal terminals. In fact, railroad terminals face similar problems when managing and optimizing the yard storage and the train (i.o., ship) load planning. The joint solution of the two tasks is, for instance, addressed in Dotoli et al. (2017). I also notice the two contributions share the baseline idea of solving the yard storage problem by solving a heuristic procedure which takes inspiration from a knapsack problem. Since only few papers address the terminal optimization as whole and as such I suggest the authors to also refer the interested readers to these contributions. In addition, several simulation-based approaches to describe the terminal dynamics as a whole also exist (see, e.g., the work in Cavone et al. (2017)). Here again, a couple of sentences in Section 1 would probably be enough.

The logical formulation of the problems seems to be correct without relevant flaws I can notice. I think that authors could realize a flowchart to better describe how the presented method works.

In my opinion the paper would gain a lot if a real case study, even if arising from some literature contribution, is presented. Currently, the method is only tested on experimental evaluations. In my view, more details on how these instances are created should be provided.

A possible improvement (which is also presented in the already suggested work in Cavone et al.) is to perform what-if analyses in the simulating environment (e.g., assuming an increase in the containers flows) and compare each other the possible alternative solutions (e.g., via multi-criteria decision making techniques). I think these possibilities may effectively support the decision making process, thus enhancing the practical relevance of the contribution. Nonetheless, this aspect is probably out of the scope of this work, hence it could only to be sketched as a future work.

Suggested references:

- M. Dotoli et al. (2017), “A Decision Support System for optimizing operations at intermodal railroad terminals”, IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 47 (3), pp. 487-501.

- G. Cavone et al. (2017), “Intermodal terminal planning by Petri Nets and Data Envelopment Analysis”, Control Engineering Practice, vol. 69, pp. 9-22

Author Response

Reviewer 1

This paper presents an integrated methodology for the joint resolution of two important tasks in port yard, namely the Container Relocation Problem (CRP) at yards and the Multi-Port Stowage Planning Problem (MPSP) for the charge and discharge of ships. More in detail, the proposed approach combines simulation, a genetic algorithm, and a rule-based heuristics, with the final aim of minimizing the total number of relocations for both the yard and the ship. The method is tested on several instances of growing complexity. Overall, results are sound and promising, as well as future research directions are clearly highlighted at the end of the paper. The paper is interesting, well written and well-structured enough. The aim of the contribution is clear, and the manuscript has a practical relevance.

Thank you for your careful reading. We addressed all points raised and carried corrections or explanations for each topic.

R.1.1. The introduction is concise, but clear and effective. Although the reference list is complete and updated enough, I think the literature review and the state of the art description could be improved. In addition, authors should better relate their work to other researches, emphasizing the paper contribution and its positioning within the existing literature, and highlighting its novelty and main advantages.

Comment: Thank you. We revised and created a new section “Related Works” section to emphasize the paper contribution.

R.1.2. In particular, I do not agree with the statement of the authors that in the relevant literature “the decisions of different interests of the diverse agents have not been incorporated”. I suggest the authors to also recall some papers dealing with the same operations when performed in multimodal and intermodal terminals. In fact, railroad terminals face similar problems when managing and optimizing the yard storage and the train (i.o., ship) load planning. The joint solution of the two tasks is, for instance, addressed in Dotoli et al. (2017).

Comment: Thank you. Indeed, the “Related Works” section mentions many papers that have integrated different container terminal related problems. We have revised our conclusion section that had that statement.

I also notice the two contributions share the baseline idea of solving the yard storage problem by solving a heuristic procedure which takes inspiration from a knapsack problem. Since only few papers address the terminal optimization as whole and as such I suggest the authors to also refer the interested readers to these contributions. In addition, several simulation-based approaches to describe the terminal dynamics as a whole also exist (see, e.g., the work in Cavone et al. (2017)). Here again, a couple of sentences in Section 1 would probably be enough.

Comment: Thank you. We added both references in the “Introduction” section with a proper comment about their contribution for the literature.

 

 

R.1.3. The logical formulation of the problems seems to be correct without relevant flaws I can notice. I think that authors could realize a flowchart to better describe how the presented method works.

Comment: Thank you. We included a new flowchart with a better description about how the method works.

R.1.4. In my opinion the paper would gain a lot if real cases study, even if arising from some literature contribution, is presented. Currently, the method is only tested on experimental evaluations. In my view, more details on how these instances are created should be provided.

Comment: Unfortunately, there are no public repositories with port data, specially the joint problem of stowage and yard planning. Although, once the paper is approved we will make available for some synthetic data about integrated problem. This data set had been created based on data of instances from papers that tackle problems separately.

R.1.5. A possible improvement (which is also presented in the already suggested work in Cavone et al.) is to perform what-if analyses in the simulating environment (e.g., assuming an increase in the containers flows) and compare each other the possible alternative solutions (e.g., via multi-criteria decision making techniques). I think these possibilities may effectively support the decision making process, thus enhancing the practical relevance of the contribution. Nonetheless, this aspect is probably out of the scope of this work, hence it could only to be sketched as a future work.

Comment: In our instances we have made variations on the distance matrix, which means how many ports the containers will be onboard, the ship, consequently altering the ship's occupancy along the route. What-if scenarios of the flow of boarding containers are an interesting possible future work, and we addressed this possibility in “Future Works” section.

Suggested references:

- M. Dotoli et al. (2017), “A Decision Support System for optimizing operations at intermodal railroad terminals”, IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 47 (3), pp. 487-501.

- G. Cavone et al. (2017), “Intermodal terminal planning by Petri Nets and Data Envelopment Analysis”, Control Engineering Practice, vol. 69, pp. 9-22

Reviewer 2 Report

The authors have proposed an integrated approach for solving The Multi-Port Stowage Planning Problem and the Container Relocation Problem. The rules for solving each of the problems are integrated into the decision rules used in the simulation. The decision rules are optimized with the use of genetic algorithm.

The authors reviewed the related research and on that basis proposed their original approach, which was experimentally verified.

The following issues should be addressed:

1. Please explain why a simple Genetic Algorithm was used for optimization of the rules - there are lots of other, more sophisticated evolutionary algorithms available.

2. Can the problem be treated as multi-objective optimization problem? If yes, maybe it would be better to apply one of the multi-objective evolutionary algorithms like NSGA or SPEA variants.

3. What is the upper limit of the number of containers for the proposed approach?

4. Can the proposed method be parallelized?

5. The English language should be improved – there are some grammar and style errors.

Author Response

Reviewer 2

The authors have proposed an integrated approach for solving The Multi-Port Stowage Planning Problem and the Container Relocation Problem. The rules for solving each of the problems are integrated into the decision rules used in the simulation. The decision rules are optimized with the use of genetic algorithm. The authors reviewed the related research and on that basis proposed their original approach, which was experimentally verified. The following issues should be addressed:

 

Thank you for your careful reading. We addressed all points raised and carried corrections or explanations for each topic.

 

R.2.1. Please explain why a simple Genetic Algorithm was used for optimization of the rules - there are lots of other, more sophisticated evolutionary algorithms available.

Comment: The Genetic Algorithm used can provide good sequences of feasible movements within a short computational time. The paper contribution is to solve an integrated problem and bring new insights of a more proper manner to do port logistics operations.  A more sophisticated evolutionary algorithm can be mentioned as future work to help the GA get out of local minimums solutions. We also believe that more intelligent rules can have positive impact on the final solution.

 

R.2.2. Can the problem be treated as multi-objective optimization problem? If yes, maybe it would be better to apply one of the multi-objective evolutionary algorithms like NSGA or SPEA variants.

Comment: No, the two problems have the same objective, which is to reduce the number of relocations. When we integrated the problems, we simply summed the number of relocations of each part of the problem and minimized it. At the end, the integrated problem has one objective. It could be multi-objective if besides the number of relocations, we were also trying to improve another objective, like the distance traveled by the containers.  We added this as a possible development in the “Future Works” section.

 

R.2.3. What is the upper limit of the number of containers for the proposed approach?

Comment: The upper limit of the number of containers for the proposed approach depends on the power of machine used. The results of the paper bring some insights of how the increase in the number of containers could affect the total computational time on a personal computer. In particular, other features like Yard and Ship Dimensions, and Yard occupancy rate impact on computational time are shown in Tables 1 and 2. Also, as suggested, improvements can be implemented as parallelize the simulation of the individuals.

 

R.2.4. Can the proposed method be parallelized?

Comment: Yes, in the simulation of each individual can be easily parallelized. We added this possibility of development in the “Future works” section.

 

R.2.5. The English language should be improved – there are some grammar and style errors.

Comment: Thank you for this observation. We made a major review to tackle all grammar and style errors.

Reviewer 3 Report

The submission solves stowage planning and yard management problem using heuristic algorithm which shows good convergence. The authors solve a very practically important problem. I think paper brings contribution to the literature. But; Following comments should be addressed to improve this work:

1. Following studies are also considering stowage planning and port yard planning. I think you can cite and discuss following papers in the related works section. 

Iris, Ç., Christensen, J., Pacino, D. and Ropke, S., 2018. Flexible ship loading problem with transfer vehicle assignment and scheduling. Transportation Research Part B: Methodological, 111, pp.113-134.

Monaco, M.F., Sammarra, M. and Sorrentino, G., 2014. The terminal-oriented ship stowage planning problem. European Journal of Operational Research, 239(1), pp.256-265.

2. Is there a way to compare your GA results to another solution method? This way, you can prove that your algorithm performs strong.

3. Did you try to solve this problem for group of containers? You may reduce problem size if you solve this problem for a set of containers going to a specific port. 

Author Response

Reviewer 3

The submission solves stowage planning and yard management problem using a heuristic algorithm which shows good convergence. The authors solve a very practically important problem. I think the paper brings a contribution to the literature. But; the Following comments should be addressed to improve this work:

Thank you for your careful reading. We addressed all points raised and carried corrections or explanations for each topic.

R.3.1. Following studies are also considering stowage planning and port yard planning. I think you can cite and discuss the following papers in the related works section. 

Iris, Ç., Christensen, J., Pacino, D. and Ropke, S., 2018. Flexible ship loading problem with transfer vehicle assignment and scheduling. Transportation Research Part B: Methodological, 111, pp.113-134.

Monaco, M.F., Sammarra, M. and Sorrentino, G., 2014. The terminal-oriented ship stowage planning problem. European Journal of Operational Research, 239(1), pp.256-265.

Comment: Thank you. We added both references in the introduction with a proper comment about their contribution to the literature.

R.3.2. Is there a way to compare your GA results to another solution method? This way, you can prove that your algorithm performs strong.

Comment: It is the first that a methodology addresses both problems with instances that are not toy problems.

R.3.3. Did you try to solve this problem for a group of containers? You may reduce problem size if you solve this problem for a set of containers going to a specific port. 

Comment: This could be an interesting heuristic strategy to solve the problem. We included this suggestion in the Future Works section. It could also be implemented as an additional rule.

Round 2

Reviewer 1 Report

I have looked at the revised version of the paper.

Authors have addressed the main remarks and comments, following the Reviewers’ suggestions and including them appropriately in the revision.

There are no more comments from my side.

Reviewer 3 Report

My comments are addressed well. Paper can be accepted.