A Model for Intact and Damage Stability Evaluation of CNG Ships during the Concept Design Stage

Round 1

Reviewer 1 Report

The proposed paper for the preliminary evaluation of design characteristics of a ship based on a regression method (in this particular case the intact and damaged stability performance of a CNG carrier) seems very interesting and promising, since the same method could be applied equally well to other design problems and/or ship types. Some comments and suggestions for the improvement of the final paper are listed in the following:

Page 3, lines 110-111: “…subject to the standard damages (Table 1) applied elsewhere or to any smaller damage…”. It seems that something is missing or mistyped here.

Page 4, lines 117-118: “In addition, the rules the IGC code defines some requirements….”. Please check syntax.

Page 5, lines 174-175: It would be useful to provide some short explanation of the CCF method, along with some reference, since it is not necessarily familiar to most readers.

Page 8, eq. (2) and eq. (3): Eq. (2) is nonlinear, while eq. (3) is linear. It would be useful if the authors can include an explanation of the derivation of eq. (3) from eq. (2).

Page 8, eq. (4): Can the authors elaborate on the derivation of eq. (4)?

Page 8, lines 210-216: The explanation of the adopted method is not quite clear. Can the authors explain further?

Please define the dashed lines in Fig. 3, 4, 5 and 6.

Page 17, lines 351-352: “LCB … is coupled only with KG/T”. It seems that LCB is also coupled with B/T. Please check.

Page 18, Table 11: The values for I3 and I4 are exactly the same. Is it really the case, or is it a typing error?

Page 19, lines 443-445: “the effective floodable lengths FL have been calculated correcting the GFL with the permeabilities as reported in Table 12”. How is the correction carried out?

Author Response

The authors would like to thank the reviewer for the corrections and suggestions given to improve paper quality. Hereunder the replies to the given comments in as per reviewer list:

(lines 110-111) the sentence has been reformulated as follows: "\Besides, the IGC ensures that the ship will be capable to survive with sufficient residual stability in case that standard damage (Table 1) occurs or a smaller one with more severe consequences." (lines 117-118) 'the rules' has been removed. A short explanation of CCF has been provided, together with additional references [22,23].    Vector X includes linear and non-linear terms of equation (2) and considers these terms as additional linear independent variables as usual in multiple linear regression. To avoid possible misunderstandings, this has been added to the text. Equation (4) is the standard result of the least square method application. This was already stated at line 205 of the original version. The elaboration on the formula requires a lot of additional space, without giving (according to authors opinion) additional information to the reader. Instead, an additional reference [25] to the elaboration has been added. Stepwise regression is a common methodology adopted to find predictors automatically. Here, the only enhancement is given by the adoption of a preliminary adimensionalisation of the data between [-1,1] to avoid the influence of the variables relative values to the regression. Additional reference to the stepwise method has been added [26]. The paragraph has been reformulated as follows: "In this study, the regressions have been performed using a stepwise selection process[26]. At each step all the variables are individually changed of status, means variables not included in the model are included, and variables inside the model are discarded. The variable whose status change (in or out of the model) is decreasing at most the sum of squared errors (SSE) value is detected and its status is flipped. If it was inside the model it is removed and vice versa. The process continues automatically until there is no variable changing the SSE over a given threshold. The process works both starting with no variables in the initial model (forward mode) or with all variables (backward mode). Here backward mode has been applied. Moreover, in order to keep the same kind of threshold value through the study, all the variables, dependent (x) and independent (y) have been normalised in [-1,1] prior to starting the regression procedure. Under these assumptions, the threshold has been set to 0.06." The lines are representative of a 95% confidence interval, it has been specified in the figures. B/T has been added to the text. Table 11 has been updated  The correction to floodable lengths has been carried out according to SOLAS 90, dividing GLF by the permeability μ provided in Table 12. This has been further specified in the text.

Reviewer 2 Report

General comments

The paper is well structured and the presented work is valuable for the initial ship design.

It is agreed that the current work presents a model for intact and damage stability evaluation of CNG ships during the concept design stage. The six hull-form parameters used in the methodology are common and independent of ship type but the generation of the database, used for the metamodels, is based on CNG ships.

Table 2 presents the range of design variables. It is worthwhile to explain why the particular ranges were selected and how realistic are these values to CNG ship design. For example, LCB varies from 3%L up to 1.5%L aftwards of midship section. In addition, B/T varies from 4.0 up to 5.0.

Conceptual design, with respect to intact and damage stability calculations, includes also a Lightweight estimation. From the reviewer’s point of view, a more comprehensive discussion is needed at least at the paragraph Conclusions/Way ahead.

Typing comments

Line 207: ” ….respectively. using ….”

Line 225 “A mentioned, ….”

Line 372 “ … other therms …”

Author Response

The authors would thank the reviewer for the comments and suggestions given to improve the paper. Hereafter the replies to comments:

The values in Table 2 are coming from previous studies of the authors [11-14] where necessities related to propulsion, station-keeping and internal layout identify these ranges for optimal vessel generation. In order to maintain compliance with the metamodels developed for other quantities, the same variants have been applied. With respect to specific stability variables (D/T, KG/T) the range has been selected in such a way to include all the possible feasible ship region. This has been specified in the text: "The ranges have been selected on the base of previous studies[11-14], where specific ranges of geometrical coefficients were identified for the development of other metamodels (e.g. seakeeping, station-keeping, propulsion and internal layout). In order to maintain compliance with the other models, the same ranges have been applied for the common variables. The specific stability variables ranges (D/T and KG/T) have been selected to investigate all the possible feasible solutions." The authors fully agree with reviewer comment. Weight estimation is for sure one of the most important issues during concept design. The MCDM process developed by the authors includes also a dedicated simplified weight estimation metamodel, providing weight and centre of mass position as a function of design variables. In the present work, the stability metamodels, have been developed with the aim to use the data coming from weight estimation as an input, without a direct coupling with stability metamodel. This is why variations have been made for KG/T. In our research program, it is planned to further investigate weight estimation, therefore, for the moment, the authors prefer to dissociate the two studies (weights and stability). This discussion has been added to the conclusions section as suggested. "It is worth to notice that the developed stability metamodels require an estimation of the ship weight and the centre of mass. In the present study, only the stability metamodels have been analysed, thus they are decoupled with ship weight estimation. Inside an MCDM process, the obtained models are structured in such a way to use data coming from a dedicated simplified ship weight estimation metamodel, and that's why the variable KG/T has been introduced. Weight estimation is planned to be analysed more in detail in further research, that may also lead to a coupling of the two issues."