Model Uncertainty for Settlement Prediction on Axially Loaded Piles in Hydraulic Fill Built in Marine Environment

Round 1

Reviewer 1 Report

The subject of uncertainty for settlement prediction on axially loaded piles in hydraulic fill built in marine environment is actual and important. Conducting experimental tests is important to confirm or reject the theory. Using statistical methods to analyse engineering issues is a very practical and sensible approach.

The article is very valuable and prepared with care, which can be seen especially in the detailed and very clear description of the graphs.

The introduction provide background and include relevant references.

The research is design and perform appropriate. All methods are described adequately. The exception is the statistical methods used.  Results are present clearly. However, there are some aspects that need to be clarified. In my opinion, it would be useful to describe the statistical methodology. For example, what methods were used to obtain the model: regression, neural networks or others? For the analysed variables there is no description of basic descriptive statistics, so it is not possible to assess whether the statistical method used is adequate or not.  For example if the distribution is not normal the choice of type of statistical forecasting methods are limited. Therefore, the statistical methodology used requires a more detailed description.

In my opinion, the analysis of the model's uncertainty should include a value to assess it. For example, the simplest MAPE assessment or other.

To sum up, I think that the article deserves attention, because it is a very well done and presented scientific work. In my opinion  the article can be published after minor additions, especially those related to statistical analysis.

Author Response

Point 1: The subject of uncertainty for settlement prediction on axially loaded piles in hydraulic fill built in marine environment is actual and important. Conducting experimental tests is important to confirm or reject the theory. Using statistical methods to analyze engineering issues is a very practical and sensible approach.

The article is very valuable and prepared with care, which can be seen especially in the detailed and very clear description of the graphs.

The introduction provide background and include relevant references.

The research is design and perform appropriate. All methods are described adequately.

Author appreciates the reviewer’s interest and the quite positive assessment.

Point 2:The exception is the statistical methods used.  Results are present clearly. However, there are some aspects that need to be clarified. In my opinion, it would be useful to describe the statistical methodology. For example, what methods were used to obtain the model: regression, neural networks or others? For the analyzed variables there is no description of basic descriptive statistics, so it is not possible to assess whether the statistical method used is adequate or not.  For example if the distribution is not normal the choice of type of statistical forecasting methods are limited. Therefore, the statistical methodology used requires a more detailed description.

In this paper, the descriptive statistics are the most elementary ones. Indeed, up to section 6, the unique introduced descriptors are arithmetic mean, standard deviation, maximum and minimum values. All of them with their simplest and well-known expressions.

Section 6: Probabilistic Approach uses the Montecarlo tecnique to solve the force-settlement model for the axially loaded pile. Montecarlo method for reliability problems is described by O. Ditlevsen et al (2007) [ref 7]. The basic description of this method request a space that authors consider is not worthy spending on this concise paper. The cited reference can be used to fully understand the procedure.

In order to implement the method, it is needed to assume a density function for the input data. In this work, authors have used the Normal Distribution Function. This point is relevant and it is clarified on line 307.

Other density function introduced in the paper is the Uniform Density Function that is already mentioned on line 357.

Authors consider that the paper provides enough information to fully develop all the performed calculations.

Point 3:In my opinion, the analysis of the model's uncertainty should include a value to assess it. For example, the simplest MAPE assessment or other.

Author fully agree that a value can be provided to assess the uncertainty of the two introduced models: the deterministic and the probabilistic one. Even though, a discussion on what should be the better descriptor and its relations with other engineering formulation would be inevitably raised.

In this paper, authors prefers to show graphically on figure 18 the raw results as they are calculated. The figure gives a clear image on the method accuracy and they may be visually compared to the measured values. On the other hand, the figure gives enough information to obtain several error indexes, allowing other research works to choose the one that suits better with their particular method.

To sum up, I think that the article deserves attention, because it is a very well done and presented scientific work. In my opinion the article can be published after minor additions, especially those related to statistical analysis.

Author again thanks the reviewer for its positive point of view.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors,

In this paper the authors present a probabilistic approach to quantify uncertainties coming from soil tests through the outputs of some pile axially-loaded tests have been review with a classic numerical model to assess pile deformation.

As weak elements:

I believe that the abstract could have been more consistent and the bibliography improved with recent references.

Only 4 of the 15 are current references (since 2010).

Please change section 7: Discursion, I think it was a mistake.

The Conclusions section is missing.

Ncoregido (Ec 2) could be renamed to English

Figures 3 and 4...The writing is distorted, please correct it.

For all terms of relations with indices please use the Equation Editor (Ex pag 10 lines 230… .235)

Ex

The solution of the previous system provides the seeking relationship between applied external pressure F0 and displacement at the top pile (u1). Since skin friction Rfi(ui) and base resistance 231 Rp(un) are functions of displacement, the system is non-linear and it has to be used an iterative 232 algorithm to reach the solution. 233

4.4. t-z and Q-z curves 234

Rfi functions 235

Please check all paper.

In Figures 11 and 12 the representation could be only on the visible area  on the ordinate Settlement (mm) val min is 0 and max could be 2.5 and not 4.

And in terms of notable elements

It is a very interesting paper that uses mathematical statistics to quantify uncertainties coming from soil tests. The authors manage through statistical tests to identify and quantify the probability with which the uncertainty of the real tests is seriously affected.

Author Response

Response to Reviewer 2 Comments

In this paper the authors present a probabilistic approach to quantify uncertainties coming from soil tests through the outputs of some pile axially-loaded tests have been review with a classic numerical model to assess pile deformation.

As weak elements:

Point 1: I believe that the abstract could have been more consistent and the bibliography improved with recent references. Only 4 of the 15 are current references (since 2010).

The paper belongs to a broader research program that studies uncertainties on geotechnical formulations. Reference section includes recent papers from this research programs. On the other hand, the paper is a review of classical and well-known formulations -both mathematical and geotechnical-, that is why author have been compelled to include classical works that are still prevalent.

Point 2: Please change section 7: Discursion, I think it was a mistake.

Corrected

Point 3:The conclusions section is missing.

Author understand that section 7 is both the study conclusions and the discussion points. None of the conclusions shall be validated until a previous discussion happens. Anyway, it seem sensible to change the tille of section 7 to Conclusion and Discussion

Point 4. Ncoregido (Ec 2) could be renamed to English

Corrected

Point 5.Figures 3 and 4...The writing is distorted, please correct it.

Corrected

Point 6. For all terms of relations with indices please use the Equation Editor (Ex pag 10 lines 230… .235)

Ex

The solution of the previous system provides the seeking relationship between applied external pressure F0 and displacement at the top pile (u1). Since skin friction Rfi(ui) and base resistance 231 Rp(un) are functions of displacement, the system is non-linear and it has to be used an iterative 232 algorithm to reach the solution. 233

4.4. t-z and Q-z curves 234

Rfi functions 235

Please check all paper.

Corrected

Point 7:In Figures 11 and 12 the representation could be only on the visible area  on the ordinate Settlement (mm) val min is 0 and max could be 2.5 and not 4.

Corrected

And in terms of notable elements

Point 8:It is a very interesting paper that uses mathematical statistics to quantify uncertainties coming from soil tests. The authors manage through statistical tests to identify and quantify the probability with which the uncertainty of the real tests is seriously affected.

Author thanks the reviewer for the disclosed interest.

Author Response File: Author Response.pdf