A Bayesian Pipe Failure Prediction for Optimizing Pipe Renewal Time in Water Distribution Networks

Round 1

Reviewer 1 Report

This study investigates the most appropriate parameters for predicting pipe failure in the optimization. In particular, the Non-Homogeneous Poisson Process (NHPP) with Markov Chain-Monte Carlo (MCMC) approach is presented for Bayesian inference, while Maximum Likelihood (ML) is applied for frequentist inference as a comparison method. The proposed method is important for failure prediction of water distribution networks. My comments are as follows:

(1)            Life-Cycle Cost (LCC) is one of the most important factors to consider when determining the most cost-effective system solution. This article applies LCC evaluation to find the optimum time to replace the pipe. Is the pipe replacement considered in the failure prediction of water distribution networks?

(2)            Is the inspection and maintenance cost considered in the LCC?

(3)            There are some obvious errors among observed, MCMC predicted and ML predicted pipe failure in specified years. It is recommended to explain it.

(4)            The application range of the proposed method is recommended to be described.

Author Response

Dear Reviewer 1 :

We really appreciate the review that was submitted. We are pleased to give feedback on the following points  :

1. In this article, pipe failure prediction models do not consider the segmented replacement of the pipe networks, which becomes the limitation of this study. We have a plan to consider it in future research.
2. Inspection and maintenance costs are not considered in the LCC. Only repair cost (Cr) is considered and it is included in the running costs (CR). This is one of the limitations of this study and the authors have added an explanation to the manuscript on page 7.
3. There are some apparent errors among the observed value, the MCMC predicted, and the ML predicted. The MCMC predicted is better than the ML predicted because it has a lower mean square error value. This article proposes some visual assessment techniques for evaluating the model concerning the resulting errors, including annual failure plots, cumulative failure plots, pair failure plots, and quartiles of pipe failures. From the results, there is a tendency of predicted failures to be consistent with the observed failures. We are delighted to be reminded to explain this, and the authors have added an explanation to the manuscript on page 17.
4. This article recommends a bayesian model for pipe prediction using the MCMC method for pipe failure modeling using the ML method as a comparison. Based on the analysis, the recommended model is appropriate for predicting failure with minimum serial data of ten years. Pipe failure predictions should be more accurate in situations with more than ten years of data series. We are pleased to be reminded to describe this and the authors have added a description to the manuscript on page 17.

Thank you for your attention. We hope that this response will be satisfactory.

Sincerely,

Authors,

Widyo Nugroho

Christiono Utomo

Nur Irawan

Reviewer 2 Report

A well presented paper with good research methodology and findings.

Author Response

Dear Reviewer 2 :

We appreciate the review that was submitted. It really helps us to improve our research capabilities.

Thank you for your attention.

Sincerely,

Authors,

Widyo Nugroho

Christiono Utomo

Nur Irawan