See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328410689

Resistance and Trim Predictions for the NPL High Speed Round Bilge
Displacement Hull Series

Conference Paper · September 1997

CITATIONS READS

14 2,944

3 authors, including:

Dejan Radojcic
University of Belgrade
81 PUBLICATIONS   203 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Tip-Driven Marine Propellers and Impellers/Obodni pogon propelera View project

Move It - Modernization of Vessels for Inland Waterway Freight Transport (2012-2014) View project

All content following this page was uploaded by Dejan Radojcic on 22 October 2018.

The user has requested enhancement of the downloaded file.

 Resistance and Trim Predictions for the NPL
High Speed Round Bilge Displacement Hull Series

by

Dejan Radojcic, Tamara Rodic, Natasa Kostic

Dept. of Naval Architecture
Faculty of Mechanical Engineering
University of Belgrade

SUMMARY

Mathematical representation of calm water resistance and trim of the systematic NPL series,
which is often used for high speed pilot boats, work boats, patrol craft, etc. is presented. A
predictive technique is established by regression analysis. Dependent variables are the resistance-
displacement ratio (RT/ for standard = 100,000 lb) and dynamic trim (), while independent
variables are length-displacement ratio (L/1/3), the ratio of length to beam (L/B) and the ratio of
beam to draught (B/T), as well as their cross-products and their different powers multiplied by
powers from 0 to 8 of the displacement Froude number (Fn) .
Similar mathematical models published previously are based on the resistance data of NPL
series combined with the data of other series. This paper analyses broader speed range (Fn = 0.8-3.0)
and is based on the NPL series only, resulting in a more reliable resistance prediction method. The
mathematical models are suitable for implementation in software and can replace the "manual"
power prediction calculations for the NPL series.

AUTHORS' BIOGRAPHIES

Dejan Radojcic is an associate professor of naval architecture at the Faculty of Mechanical

Engineering, University of Belgrade. From the same Faculty he received his Dipl. Ing., M.Sc. and
Ph.D. degrees (all concerning high speed craft) in 1975, 1982 and 1987, respectively. Following
graduation he was employed for three years in two local shipyards. He spent one year at the Ship
Science Department, University of Southampton, as a British Council scholar. He is a member of
RINA and SNAME, and currently is a president of the Society of Naval Architects and Technicians
"Belgrade".

Tamara Rodic graduated with distinction from the faculty of Mechanical Engineering,
University of Belgrade in 1995 at which time she received her Dipl. Ing. degree in naval
architecture. Currently she is a post graduate student at the Department of Naval Architecture,
Faculty of Mechanical Engineering. She is a member of the Society of Naval Architects and
Technicians "Belgrade".

Natasa Kostic graduated with distinction from the faculty of Mechanical Engineering,
University of Belgrade in 1995 at which time she received her Dipl. Ing. degree in naval
architecture. Currently she is a post graduate student at the Department of Naval Architecture,
Faculty of Mechanical Engineering. She is a member of the Society of Naval Architects and
Technicians "Belgrade".
1. INTRODUCTION

The NPL systematic high speed round bilge displacement (for higher speeds
semidisplacement) hull series, published more than 20 years ago [1], is well known and is still rated
as the most useful series for work boats, patrol craft, pilot boats, etc. It is designed for operation in
the Froude number range FnL= 0.3-1.2 (Fn = 0.6-3.0). The series covers :
- length-beam ratio L/B = 3.33-7.50
- length-displacement ratio (M) = 4.5-8.3
- beam-draft ratio B/T = 1.75-10.77
while constant values are taken for :
- position of longitudinal center of buoyancy LCB = 6.4%L aft amidships
- block coefficient CB = 0.397
- ratio of transom area to maximum section area AT/AX = 0.52.
Besides calm water resistance and dynamic trim, Reference [1] enables calculation of maneuvering
and seakeeping characteristics, stability underway, propulsion coefficients, rise and fall of CG, as
well as the influence of transom wedge and spray rail.
Other published relevant round bilge series are SKLAD, Series 63, SSPA series etc. The
advantages and reliability of the NPL series has made it a standard for small craft. Many authors
have used the NPL body lines and characteristics to develop new series, as in for example
Reference [2].
It is surprising that although it is so widely used, a literature search did not turn up any
evidence of a mathematical representation of the resistance and trim data specific to the NPL series.
That is, mathematical models based on the resistance data of the NPL series combined with the data
of other series are available. In the Mercier and Savitsky [3] model, regression analysis was applied
to the resistance data of 7 transom-stern hull series (NPL, Nordstrom, DeGroot, SSPA, Series 64,
Series 63 and Series 62), which included 118 separate hull forms. In the Finnish VTT [2] model,
resistance prediction equation was developed using the regression analysis based on the NPL, SSPA
and VTT series, as well as on some other suitable models from the records; 65 models in all. Both
resistance equations were developed for total resistance-displacement weight ratio RT/ for
standard displacement of 100,000 lb (45.36 t). The Mercier and Savitsky model is suggested for use
with lower speeds (Fn between 1 and 2) while the Finnish VTT model is suitable for higher speeds
(Fn between 1.8 and 3.2). In both cases, the NPL data are one among many other data sets, so that
the resistance prediction equations are not very reliable when used specifically for the NPL hull
forms.
Consequently, this paper is directed towards mathematical representation of the resistance and
dynamic trim specifically for the NPL series. Polynomials derived to describe (RT/)100000 and 
enable application of existing computer routines, thereby replacing manual calculations.
Representation of dynamic trim is necessary for oblique flow propulsor (propeller) evaluation. This
enables feasibility studies with power as an objective function (min Pd), rather than having to
perform separate minimization of resistance and maximization of propulsor efficiency, as is usually
done. This approach (min Pd) and its advantages are given in the Reference [4].

2. REGRESSION ANALYSIS TECHNIQUES - AN OVERVIEW

Regression analysis has been successfully used to analyze the resistance data for both,
random hull forms, and methodical series. The equations based on random hull forms have broader
applicability but are often not reliable. The reliability aspect of the model is very important,
particularly for application in everyday problems when the correct value is not known in advance.
The individual characteristics of each series, i.e. the secondary hull form parameters, can not be
taken into consideration when several random hull forms are treated simultaneously. Secondary hull
form parameters may be lost among primary parameters, even if many independent (explanatory)
variables are introduced into the mathematical model. Therefore, it is often better to narrow the
applicability to particular methodical series, and so to increase the reliability of the model. With this
approach only the primary hull form parameters are considered, but the user, having in mind the
subject hull, should be familiar with the other characteristics of the series which are not explicitly
included in this mathematical model. The disadvantage of this approach is that multiple models are
required, each for given methodical series or a group of very similar hull forms, instead of having
only one model, or few models needed for random hull form approach.
Various resistance models were used, from those based on the wave theory to the ones using
brute force regression analysis with exponential, logarithmic, polynomial or some other functions.
It has been found that, when the number of terms is relatively high and when the cross-products and
different powers of the independent variables are used, there is no need for function
transformations. That is, the original polynomial form sufficed.
Reference [5] gives a very good summary of the history of use of regression equations for
resistance evaluation. Two general types of regression equations for resistance evaluation have
evolved:
- Speed-independent models, when separate equations are generated for a series of discrete speeds,
since the speed is not included as an independent variable.
- Speed-dependent models, with vessel's speed included as an independent variable.
The advocates of speed-dependent models claim that the predicted resistance often does not
vary properly with speed, since the resistance computed at one speed is not directly linked to that at
another speed. This is because the speed variable is not explicitly included in the regression, as
explained in, for example, References [5] and [6]. The accuracy of the speed-independent models
is, believed to be, somewhat better since independent equations are developed for each speed.
Reference [7] offers a compromise solution that has been used by the first author on several
occasions. Note that this method is similar to the approach given in Reference [8]. Essentially, with
this method the speed-independent equations (with the same variables for whole speed range) are
first developed. A second regression analysis is then performed with the regression coefficients
cross-faired against speed (or Froude number). Thus, accurate speed-independent equations are
obtained for discrete Froude numbers through the first step; the second step provides speed-
dependent equation. Of course, either of equations may be used independently to estimate
resistance.
An important and delicate part of this method is the formation of the "best subsets" from the
initial (for all speeds the same) equation. The usual statistics (coefficient of determination, t-test or
F-test, standard deviation, significance test for each variable etc.) were found to be insufficient and
in fact were sometimes misleading. Therefore, trial and error technique was also used to define the
best subsets for the whole speed range. Some variables, judged to be less significant were rejected
deliberately, although a stepwise method was used throughout the analysis. The accuracy of the
model, a primary goal of this analysis, was checked after every incremental step in the process was
undertaken. It should be pointed out, however, that several very good, although dissimilar, models
could be derived.

3. SURVEY OF THE MATHEMATICAL MODELS

APPLIED TO HIGH SPEED CRAFT

The aforementioned Mercier and Savitsky [3] is the speed-independent mathematical model,
while the one for the VTT series [2] is the speed-dependent model. Jin's speed-independent model,
described in Reference [9], for calculation of residuary resistance of the round bilge displacement
hulls over a speed range of FnL = 0.4-1.0, should also be mentioned. Speed-independent and speed-
dependent models are given in References [5] and [6], respectively for the high speed transom stern
ships (FnL = 0.15-0.9).
 Reference [10] gives information about high speed displacement hull series AMECRC (which
is an extension of MARIN series) and is in many respects similar to the NPL series. A nonlinear
speed-independent method was used there, since it was proven to be better than the least square
minimization used by multiple regression analysis. Unfortunately, mathematical model itself was
not given.
Resistance and trim of planing hulls for Fn = 1.0-3.0 can be estimated from the speed-
independent and speed-dependent mathematical models given in Reference [11], while speed-
independent model for resistance prediction for the hard chine catamaran hull series '89
(Fn = 1.0-3.5) is given in the Reference [12].
For the sake of completeness, two more references concerning the NPL series should be
mentioned. Reference [13] deals with the propulsion coefficients of the NPL series, which are
necessary for power estimation. Reference [14] deals with the method for the determination of
hydrostatic data and form stability characteristics for the NPL series, but doesn't have any
connections to the regression analysis.
An excellent background and guidance to designers regarding the application of numerical
methods through the use of commercially or individually developed computer software may be
found in Reference [15].

4. DEVELOPMENT OF THE PREDICTION EQUATIONS

4.1. INITIAL MATHEMATICAL MODEL

Residuary resistance-displacement ratio (RR/ [kN/t]) and running trim ( [o]) data, drawn
against (M) = L/1/3 and Fn are presented graphically in Reference [1] for five L/B groups of
models (L/B = 3.33, 4.55, 5.41, 6.25 and 7.5). Model-size wetted area (Lm = 2.54 m) at rest is
drawn as a function of (M) and L/B only. These diagrams were scanned and transformed to
numerical form forming between 120 and 160 data points for each group of the L/B models.
Breadth-draft ratio could now be calculated, since B/T = (CB  (M)3 ) / (L/B)2, (CB = 0.397 =
const.).
Three principal hull form and loading parameters were chosen for further evaluation, covering
the following range:

3.33 < L/B < 7.50 4.50 < (M) < 8.30 1.76 < B/T < 10.77.

These parameters were transformed into another set of variables with a range from -1 to +1. The
new variables, which were subsequently used throughout the regression analysis, are:

x1 = (L/B - 5.415) / 2.085

x2 = ((M) - 6.4) / 1.9

x3 = (B/T - 6.2615) / 4.5063.

(RT/)100000 (for  = 100,000 lb = 45.36 t), o and (S) = S/2/3 were chosen as the dependent
variables. Transformation from given residuary resistance and model-size wetted area format to the
new format, that was used throughout the analysis, was done according to the flow chart given in
Figure 1.
 1.026 t m 3 g9.81m s2  1.1883 10 6 m s
2
KNOWN NEW
VARIABLES
 100 ,000 lb 45.36 t g 445 kN 44.21 m 3 VARIABLES

 
1 
Fn  [-] v  Fn g 3
 vL 0.075
 Rn  CF 
1
  log R n 2 2
(M) [-] L  (M)  3

2
 L  1
S Sm  R F   C F S v2
2
Sm [m ] 
 2.54  2

RR  kN  RR RT  R
 t  R T R F      T
       100000 g
 S
( S) 
 
2
3


 [] 

Fig. 1 Transformation of resistance and wetted area from one format to another

Polynomial form was chosen for all three mathematical models. The initial polynomial
equation, which was subsequently used for the speed-independent least square curve fitting, had 27
terms, i.e.

(RT/)100000, o, (S) = a0 + a1x1 + a2x2 + a3x3 + a4x12 + a5x22 + a6x32 + a7x1x2 + a8x1x3 + a9x2x3 + a10x12x2 +

+ a11x12x3 + a12x22x1 + a13x22x3 + a14x32x1 + a15x32x2 + a16x13 + a17x23 + a18x33 + a19x13x2 +

+ a20x13x3 + a21x23x1 + a22x23x3 + a23x33x1 + a24x33x2 + a25x12x22 + a26x12x32 + a27x22x32.

From now on x4 = x12, x5 = x22 ... x15 = x32x2 ... x27 = x22x32.

4.2. SPEED-INDEPENDENT MATHEMATICAL MODELS

The 27 term polynomial equation, defined above, was the starting point for all calculations
presented in this paper, i.e. for the 12 speed-independent resistance and dynamic trim equations
(one for each Fn, Fn = 0.8, 1.0, 1.2 ... 3.0), and 1 wetted surface coefficient equation (which
doesn't depend on speed). All further statistical calculations were done with a program Statistica 5.0
(StatSoft Inc.) and specifically the subroutine for forward stepwise regression.
Several models, actually more than 600 instances concerning the resistance alone, were
obtained following the procedures which have been briefly described above. However,
representation for the lower Froude numbers was always relatively poor, so Fn = 0.6 was rejected
from further consideration. Therefore, speed-independent models are valid for Fn range between
0.8 and 3.0, while speed-dependent models are valid for Fn = 1.0-3.0, due to instability between
Fn = 0.8-1.0.
The final speed-independent resistance, dynamic trim and wetted surface polynomial terms,
regression coefficients and control information are given in the Appendix 1, Tables 1, 2 and 3,
respectively for (RT/)100000 ("a" regression coefficients), o ("b" regression coefficients) and (S)
("c" regression coefficients).
4.3. SPEED-DEPENDENT MATHEMATICAL MODELS

The speed-dependent models for resistance and dynamic trim were developed through cross-
faring of resistance and dynamic trim regression coefficients against Fn , i.e. through an evaluation
of ai = f (Fn) and bi = f (Fn). Different powers of Fn were tried, and satisfactory results were
obtained when Fn was raised to sixth power for lower Fn and eight power for higher values of Fn.
It was felt that higher Froude numbers are more important, so both ai = f (Fn) and bi = f (Fn) are
eight order polynomial equations.
Of course, transformation of Fn was introduced as well, so the new n variable, ranging
from -1 to +1, was subsequently used throughout the speed-dependent mathematical models

n = (Fn - 1.8) / 1.2.

New regression coefficients together with their control information are given in the Appendix
2, Tables 4 and 5 respectively for resistance ("" regression coefficients) and dynamic trim (""
regression coefficients). The final speed-dependent models for resistance and dynamic trim have the
following form:

 RT  8
 8   8 
     i  n i     i 9  n i x 1       i 117  n i x 27
  100000 i 0  i 0   i 0 

8
 8   8 
   i  n i    i 9  n i  x 2      i 126  n i x 27 .
i 0  i 0   i 0 

If the displacement of the new design (subject hull) is approximately 45 t, than there is no
need for wetted area evaluation. Otherwise the wetted area coefficient (obtained in the previous
step, Table 3) can be evaluated from the following equation:

(S) c 0  c1 x 1    c 26 x 26 .

The step-by-step calculation procedure for arbitrary size of the subject hull is given in the
Reference [11], for instance.

4.4. BOUNDARIES OF APPLICABILITY

Boundaries of applicability are very important part of the mathematical model, since "diligent
but stupid" computers are intended to be used. The mathematical model presented should only be
used to predict the performance of a new hull whose characteristics (secondary hull form
parameters as well) are similar to the data underlying its derivation, i.e. to the NPL series. The
distribution of principal parameters should be as shown in Figure 2 (feasible parameters should be
inside the odd-shaped box). All nine boundaries (surfaces) are relatively simple and are equations
of inequality type. This approach enables application of nonlinear constrained optimization
routines, as is discussed in the Reference [4].
 a ( M )  4.5 f  ( L / B)  3.33
b ( M )  0.6757  ( L / B)  14324
. g Fn  0.8
c ( L / B)  7.5 h Fn  3.0
d ( M )  8.3 i Fn  0.714  ( M )  0103
.  ( L / B)  0.872
e ( M )  01534
.  ( L / B) 3  19906
.  ( L / B) 2  8.732  ( L / B)  5.9683

Fig. 2 Boundaries of applicability of mathematical models

5. ACCURACY OF THE PREDICTED MATHEMATICAL MODELS

The quality of the models, i.e. important statistics like standard error, coefficient of
determination, F-test etc. are given for each regression fitted equation and are shown in the Tables 1
to 5. These statistics are not important for the end user, but are good measures of the model validity.
The degree of agreement for resistance and dynamic trim, between the NPL series and the results
obtained by the proposed models, are illustrated in the Figures 3 to 8 in the Appendix 3. The
comparison is satisfactory even for speed-dependent models.
The discrepancies for all speed-independent and speed-dependent models are summarized in
the Table 6, Appendix 3. For instance, for towing power and  = 45.36 t out of 678 data points used
in the model derivation, only 6 low-Froude-number-points (Fn = 0.8 and 1.0) "jumped over" 5%
error barrier, none of them above 7%.
A very important part of the accuracy checking is identification of any instabilities which may
occur between points used in model derivation. Therefore, an examination of the intermediate
values of principal parameters, which were not employed in the development of the mathematical
model, is necessary; see for instance Figures 9 to 12, Appendix 4.

6. CONCLUDING REMARKS

For the well-known NPL series, three groups of mathematical models were derived, speed-
independent and speed-dependent models for (RT/)100000 and , and one for (S) evaluation. They
can be used together with other available performance prediction methods as given in, for instance,
References [2] through [6] and [9] through [12]. The validity of the model across a relatively wide
speed range should be pointed out (Fn = 0.8-3.0), especially in the relation to the currently
available models presented in the References [3] and [2], valid for Fn = 1.0-2.0 and Fn = 1.8-3.2,
respectively.
Mathematical models are presented in the usual format which facilitates the application of
existing computer routines. This was one of the original goals of this study. Resistance and trim
equations enable power performance evaluation as given in Reference [4].
Moreover, it should be noted that tank testing is rarely used in design of the small craft due to
the high price of the tests in relation to the cost of the vessel. However, numerical towing tank
performance predictions, whose mathematical models for resistance and trim evaluations for the
NPL series are presented here, are often used in all design phases. The mathematical models used
for small craft performance evaluations, therefore, need to be both reliable and accurate. Both,
speed-independent and speed-dependent models presented here meet these criteria.

NOMENCLATURE

AT m2 Transom area
AX m2 Maximum section area
B m Breadth (beam) of hull on DWL
CB - Block coefficient
CF - Frictional resistance coefficient
CR - Residuary resistance coefficient
DWL - Designed waterline
FnL - Froude number (v/gL)
Fn - Volumetric Froude number (v/g1/3)
g m/s2 Acceleration of gravity (9.81)
L m Length on DWL
LCB %L Longitudinal centre of buoyancy
(M) - Length-displacement ratio (L/1/3)
Pd kW Power delivered to the propeller
Rn - Reynolds number (vL/)
RF kN Frictional resistance
RR kN Residuary resistance
RT kN Total resistance of bare hull
S m2 Wetted surface
(S) - Wetted surface coefficient (S/2/3)
T m Draught at DWL
v m/s Speed
 t Displacement mass
 m3 Displacement volume
 t/m3 Mass density of salt water (1.026)
 m2/s Kinematic viscosity of salt water (1.188310-6 at 15oC)
 o
Dynamic (running) trim angle
REFERENCES

1. Bailey, D. - "The NPL High Speed Round Bilge Displacement Hull Series", Maritime
Technology Monograph No. 4, RINA, 1976.

2. Lahtiharju, E., Karppinen, T., Hellevaara, M. and Aitta, T - "Resistance and Seakeeping
Characteristics of Fast Transom Stern Hulls with Systematically Varied Form", Trans.
SNAME, Vol. 99, 1991.

3. Mercier, J.A. and Savitsky, D. - "Resistance of Transom-Stern Craft in the Pre-Planing

Regime", Davidson Laboratory Report SIT-DL-73-1667, June 1973.

4. Radojcic, D. - "An Engineering Approach to Predicting the Hydrodynamic Performance of

Planing Craft Using Computer Techniques", Trans. RINA, Vol. 133, 1991.

5. Fung, S.C. - "Resistance and Powering Predictions for Transom Stern Hull Forms During
Early Stage Ship Design", Trans. SNAME, Vol. 99, 1991.

6. Fung, S.C. and Liebman, L. - "Revised Speed-Dependent Powering Predictions for High
Speed Transom Stern Hull Forms", Proc. Third Int. Conf. FAST '95, Travemunde, Sept. 1995.

7. Radojcic, D. - "Mathematical Model for Evaluation of Some Hydrodynamic Characteristics of

Planing Hulls", Ph.D Thesis, University of Belgrade, 1987 (in Serbian).

8. Swift, P.M., Nowacki, H. and Fischer, J.P. - "Estimation of Great Lakes Bulk Carrier
Resistance Based on Model Test Data Regression", Marine Technology, Oct. 1973.

9. Jin, P., Su, B. and Tan, Z. - "A Parametric Study on High-Speed Round Bilge Displacement
Hulls", High Speed Surface Craft, Sept. 1980.

10. Bojovic, P. - "Resistance of AMECRC Systematic Series of High-Speed Displacement Hull

Forms", NAV & HSMV Int. Conf., Sorrento, March, 1997.

11. Radojcic, D. - "An Approximate Method for Calculation of Resistance and Trim of the
Planing Hulls", SNAME Symp. on Powerboats, Sept. 1985. Also University of Southampton,
Ship Science Report No. 23, Dec. 1985.

12. Zips, J.M. - "Numerical Resistance Prediction Based on the Results of the VWS Hard Chine
Catamaran Hull Series '89", Int. Conf. FAST '95, Travemunde, Sept. 1995.

13. Bailey, D. - "A Statistical Analysis of Propulsion Data Obtained from Models of High Speed
Round Bilge Hulls", RINA Symp. on Small Fast Warships and Security Vessels, London,
1982.

14. Cassella, P. and Krauss, R. - "Caratteristiche geometriche e di stabilita delle carene della serie
N.P.L.", Tecnica Italiana, N. 3, 1981.

15. MacPherson, D. M. - "Reliable Performance Prediction Techniques Using a Personal

Computer", Marine Technology, Oct. 1993.
APPENDIX 1 - SPEED-INDEPENDENT MODELS

 RT 
   a 0 a 1 x 1 a 2 x 2 a 3 x 3 a 5 x 5 a 12 x 12 a 13 x 13 a 16 x 16 a 19 x 19 a 20 x 20 a 22 x 22 a 23 x 23 a 26 x 26 a 27 x 27
   100000

Fn 
ai 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
a0 0.012677 0.031092 0.057789 0.070863 0.080384 0.092592 0.105658 0.113350 0.118892 0.123105 0.120859 0.115058
a1 -0.008102 0.000000 0.008445 0.004134 0.000962 0.005733 0.010769 0.007299 -0.001703 -0.014977 -0.041899 -0.077104
a2 0.000061 -0.014184 -0.042841 -0.044967 -0.041259 -0.043200 -0.046379 -0.042913 -0.039950 -0.034003 -0.017179 0.005255
a3 -0.002725 0.007983 0.015331 0.013614 0.014880 0.022656 0.031878 0.031266 0.023441 0.008990 -0.022888 -0.067378
a5 -0.001185 0.002904 0.010337 0.013687 0.009803 0.008203 0.007646 0.013292 0.026694 0.043975 0.057718 0.071086
a12 0.004254 -0.003601 0.000000 0.005111 0.012468 0.012205 0.010466 0.000922 -0.018860 -0.043354 -0.046133 -0.048406
a13 -0.025702 0.000000 -0.000655 -0.013832 -0.038512 -0.043373 -0.050962 -0.050318 -0.036356 0.000000 0.051148 0.057769
a16 0.001395 0.000062 0.000000 0.002578 0.004118 0.005143 0.006964 0.008068 0.009177 0.009824 0.009804 0.008532
a19 0.004913 0.005626 0.004361 0.004098 0.004899 0.004763 0.010023 0.012955 0.012276 0.006514 -0.008769 -0.034102
a20 -0.005901 0.002059 0.010917 0.001183 0.000667 0.002649 0.002493 -0.003659 -0.018867 -0.033644 -0.057179 -0.101116
a22 0.015476 0.000000 -0.003984 0.011616 0.030670 0.033934 0.044572 0.047574 0.025806 -0.015288 -0.047272 -0.062657
a23 -0.008887 0.002059 0.006095 0.001874 -0.004834 -0.006093 -0.009346 -0.010929 -0.015162 -0.020789 -0.019478 -0.051664
a26 -0.008427 0.005375 0.016608 0.006989 0.001094 -0.001804 -0.010127 -0.017531 -0.033868 -0.050023 -0.060285 -0.118198
a27 -0.033521 0.010811 0.008090 -0.023857 -0.057101 -0.064610 -0.081529 -0.080688 -0.063470 -0.014102 0.040157 0.073258
2
R 0.9847 0.9950 0.9977 0.9966 0.9952 0.9944 0.9931 0.9946 0.9952 0.9902 0.9952 0.9985
F-test 232.5 1001.5 1963.0 1064.9 741.9 647.8 519.1 636.2 665.3 319.9 525.9 1209.7
Std. 0.00045 0.00078 0.00109 0.00136 0.00154 0.00153 0.00168 0.00151 0.00140 0.00209 0.00142 0.00084
err.
Number
61 61 61 61 61 61 61 59 56 51 47 38
of cases

Table 1 Resistance polynomial terms, regression coefficients and important statistics

  b 0  b 2 x 2  b 7 x 7  b 9 x 9  b 10 x 10  b 11 x 11  b 14 x 14  b 18 x 18  b 19 x 19  b 20 x 20  b 23 x 23  b 24 x 24  b 25 x 25  b 26 x 26  b 27 x 27

Fn 
bi 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
b0 0.175770 0.425175 0.883371 1.586455 1.899628 2.026007 2.058107 2.054187 2.069412 2.139490 2.357645 2.804328
b2 -0.305797 -0.267412 -0.418518 -0.925061 -0.732979 -0.429611 -0.236105 -0.159138 -0.247707 -0.519239 -1.217875 -2.230391
b7 0.320664 0.041130 0.000000 0.042296 -0.272631 -0.472295 -0.567934 -0.598107 -0.404230 0.067575 1.151239 2.370417
b9 0.176390 0.241145 0.000000 -0.084784 -0.058048 0.161832 0.268277 0.368525 0.669915 1.201438 1.930230 2.947236
b10 -0.168558 -0.934109 -2.238618 -1.883572 -1.691190 -1.353073 -0.947598 -0.708352 -0.347415 0.178509 1.221190 2.742371
b11 -0.103548 -0.954868 -2.517245 -2.251547 -1.890948 -1.075607 -0.193149 0.622003 1.787335 3.090234 5.035156 6.999101
b14 0.349273 -0.937388 -4.491429 -4.304975 -3.724361 -2.552354 -1.283559 -0.064981 1.761040 3.810824 7.157361 11.191026
b18 0.326051 -0.038224 -1.623344 -1.807736 -1.671105 -1.377836 -0.998759 -0.527308 0.283465 1.353872 3.053972 5.423357
b19 -1.040794 -0.806155 -0.703548 -1.350820 -0.537518 0.572009 1.628775 2.391011 3.030234 3.601990 3.710648 3.494198
b20 -0.383581 -1.348226 -2.781868 -2.624578 -2.221716 -1.349497 -0.344831 0.462487 1.541927 2.711225 4.260309 5.970641
b23 -0.078395 -0.557945 -2.635119 -2.260988 -1.924576 -1.229068 -0.401107 0.224152 1.045894 1.831750 3.596974 5.946926
b24 -0.358077 -0.434711 -1.523450 -2.023459 -1.530763 -1.018361 -0.408611 -0.118446 0.089195 0.194697 0.643298 1.373248
b25 0.794836 0.941649 1.417955 1.829837 1.260767 0.440716 -0.394656 -0.923245 -1.456637 -2.136660 -2.931875 -3.776398
b26 -0.342870 -1.516333 -4.571710 -4.392841 -3.747926 -2.234404 -0.422111 1.016714 2.912691 4.895521 8.005211 11.584463
b27 0.467308 0.865865 1.315406 1.042975 0.496254 -0.048087 -0.516796 -0.524839 -0.360649 -0.141175 -0.179117 -0.382556
2
R 0.9702 0.9656 0.9816 0.9858 0.9866 0.9881 0.9916 0.9938 0.9955 0.9965 0.9965 0.9948
F-test 83.6 72.1 169.2 178.4 188.7 212.8 305.3 414.3 564.7 734.0 741.9 496.4
Std. 0.0192 0.0813 0.1397 0.1386 0.1186 0.0935 0.0687 0.0569 0.0541 0.0573 0.0701 0.1038
err.
Number
51 51 51 51 51 51 51 51 51 51 51 51
of cases

Table 2 Dynamic trim polynomial terms, regression coefficients and important statistics
(S)  c 0  c1 x 1  c 2 x 2  c 3 x 3  c 5 x 5  c 6 x 6  c10 x 10  c15 x 15  c18 x 18  c19 x 19  c 22 x 22  c 24 x 24  c 26 x 26

c0 6.699962
c1 -2.538538
c2 3.615313
c3 0.513948
c5 0.071497
c6 1.172089
c10 -0.427145
c15 2.249989
c18 -1.769779
c19 -0.125797
c22 -0.631931
c24 1.501974
c26 0.342678

R2 0.9999
F-test 29467.7
Std. 0.0143
err.
Number
61
of cases
Table 3 Wetted surface polynomial terms, regression coefficients and important statistics
 RT  8
 8   8   8   8   8   8 
    i  n i   i  9  n i x 1   i 18  n i x 2   i  27  n i x 3   i  36  n i x 5   i  45  n i x 12   i  54  n i x 13 
   100000 i  0  i 0   i 0   i 0   i 0   i 0   i 0 
 8   8   8   8   8   8   8 
  i  63  n i x 16   i  72  n i x 19   i 81  n i x 20   i  90  n i x 22   i  99  n i x 23   i 108  n i x 26   i 117  n i x 27
 i 0   i 0   i 0   i 0   i 0   i 0   i 0 

no=1 n1=n n2=n2 n3=n3 n4=n4 n5=n5 n6=n6 n7=n7 n8=n8 R2 F-test Std. error Number of cases
0 0.093013 1 0.079344 2 -0.009610 3 -0.198929 4 0.053610 5 0.640899 6 -0.466529 7 -0.553399 8 0.476649 0.9999 8740.0 0.00047 12
9 0.006121 10 0.036347 11 -0.019097 12 -0.364107 13 0.197922 14 0.881361 15 -0.805230 16 -0.683057 17 0.672619 0.9998 1962.1 0.00068 12
18 -0.043443 19 -0.021043 20 -0.001592 21 0.317451 22 -0.138387 23 -1.126365 24 0.951995 25 0.995728 26 -0.929093 0.9997 1136.1 0.00065 12
27 0.023486 28 0.057679 29 -0.023726 30 -0.348106 31 0.179199 32 0.737798 33 -0.707577 34 -0.548689 35 0.562535 0.9997 1266.3 0.00091 12
36 0.007789 37 -0.012854 38 0.048101 39 0.128945 40 0.017943 41 -0.038860 42 -0.182194 43 -0.067477 44 0.169689 0.9997 1138.5 0.00081 12
45 0.012397 46 0.000699 47 -0.041454 48 -0.078758 49 -0.411017 50 -0.020603 51 1.004690 52 0.100286 53 -0.614613 0.9989 337.9 0.00150 12
54 -0.045620 55 -0.043439 56 0.100568 57 -0.136304 58 0.117069 59 0.787725 60 -0.349993 61 -0.584287 62 0.212039 0.9980 184.2 0.00316 12
63 0.005396 64 0.006646 65 0.001613 66 0.027995 67 -0.038633 68 -0.088732 69 0.106999 70 0.065021 71 -0.077782 0.9979 175.8 0.00033 12
72 0.005845 73 0.015480 74 0.037527 75 -0.014682 76 -0.141940 77 -0.070899 78 0.174299 79 0.062046 80 -0.101799 0.9985 247.4 0.00095 12
81 0.002700 82 0.023412 83 -0.052987 84 -0.415022 85 0.269027 86 1.022733 87 -0.841006 88 -0.773952 89 0.663990 0.9994 588.2 0.00158 12
90 0.035937 91 0.044983 92 0.032679 93 0.147806 94 -0.993943 95 -1.042473 96 2.219790 97 0.948219 98 -1.455671 0.9987 289.4 0.00239 12
99 -0.007521 100 -0.010052 101 0.052716 102 -0.110748 103 -0.257833 104 0.301729 105 0.394981 106 -0.194182 107 -0.220723 0.9980 186.6 0.00130 12
108 -0.003659 109 -0.016928 110 0.006569 111 -0.297305 112 -0.143829 113 0.808866 114 0.059827 115 -0.591199 116 0.059515 0.9988 321.5 0.00249 12
117 -0.068471 118 -0.066821 119 0.061252 120 -0.300738 121 0.829897 122 1.379352 123 -1.966275 124 -1.031222 125 1.236290 0.9985 245.7 0.00369 12

Table 4 Resistance regression coefficients and their important statistics

 8
 8   8   8   8   8   8   8 
   i  n i   i  9  n i x 2   i 18  n i x 7   i  27  n i x 9   i  36  n i x 10   i  45  n i x 11   i  54  n i x 14   i  63  n i x 18 
i 0  i 0   i 0   i 0   i 0   i 0   i 0   i 0 
 8   8   8   8   8   8   8 
  i  72  n i x 19   i 81  n i x 20   i  90  n i x 23   i  99  n i x 24   i 108  n i x 25   i 117  n i x 26   i 126  n i x 27
 i 0   i 0   i 0   i 0   i 0   i 0   i 0 

no=1 n1=n n2=n2 n3=n3 n4=n4 n5=n5 n6=n6 n7=n7 n8=n8 R2 F-test Std. error Number of cases
0 2.033088 1 0.127322 2 -1.698102 3 7.385583 4 -5.298216 5 -15.392440 6 17.085345 7 10.387316 8 -11.826407 0.9993 565.3 0.0394 12
9 -0.445368 10 2.163424 11 -2.043204 12 -13.114889 13 17.699596 14 25.430458 15 -40.013487 16 -17.534173 17 25.628880 0.9950 74.2 0.0805 12
18 -0.474875 19 -1.209746 20 2.486439 21 3.532062 22 -9.678590 23 -1.985603 24 19.995371 25 0.638182 26 -10.933303 0.9990 384.6 0.0508 12
27 0.143585 28 0.932906 29 -1.156282 30 -2.255990 31 11.399220 32 4.973652 33 -16.620768 34 -2.129468 35 7.660419 0.9997 1358.4 0.0296 12
36 -1.313364 37 1.397767 38 -0.172147 39 10.077473 40 -5.904343 41 -43.222770 42 42.104698 43 39.686211 44 -39.913942 0.9973 136.9 0.1402 12
45 -1.074402 46 4.285762 47 1.742878 48 7.116343 49 -1.468058 50 -39.316588 51 35.581547 52 38.278200 53 -38.148097 0.9994 607.5 0.1423 12
54 -2.538026 55 5.789688 56 1.867719 57 19.736459 58 -1.220753 59 -97.746012 60 76.166134 61 94.757031 62 -85.623977 0.9993 569.9 0.2349 12
63 -1.365006 64 1.504390 65 0.933899 66 9.069456 67 3.835651 68 -40.422098 69 23.516788 70 39.803665 71 -31.454296 0.9998 1555.6 0.0647 12
72 0.569334 73 7.844966 74 -2.930311 75 -28.830351 76 37.633001 77 61.195760 78 -84.867294 79 -43.279656 80 56.161719 0.9984 240.1 0.1527 12
81 -1.323953 82 5.036429 83 1.245051 84 1.417878 85 4.007709 86 -24.916260 87 17.637162 88 26.233226 89 -23.368297 0.9994 667.4 0.1268 12
90 -1.200749 91 3.403904 92 1.853146 93 13.498226 94 -13.727450 95 -68.020821 96 72.651723 97 66.400207 98 -68.913814 0.9985 246.6 0.1875 12
99 -0.977243 100 3.779033 101 -1.710287 102 -8.874362 103 12.493018 104 -1.935102 105 -6.622121 106 11.688065 107 -6.466723 0.9984 239.8 0.0738 12
108 0.419615 109 -5.617447 110 2.168412 111 17.290572 112 -24.321826 113 -29.826274 114 42.437218 115 17.202880 116 -23.530578 0.9997 1194.7 0.0624 12
117 -2.187318 118 8.724710 119 3.483912 120 5.944942 121 -1.689095 122 -63.179692 123 59.375699 124 67.577259 125 -66.469013 0.9993 568.6 0.2494 12
126 -0.087033 127 -3.018830 128 3.801864 129 4.991406 130 -5.547206 131 3.758720 132 -5.683562 133 -7.913331 134 9.316142 0.9986 277.4 0.0453 12

Table 5 Dynamic trim regression coefficients and their important statistics

APPENDIX 3 - THE DEGREE OF AGREEMENT FOR THE DATA
EMPLOYED IN THE DEVELOPMENT OF THE MODELS

1,2

actual
speed-dependent
1,0 speed-independent
+
RR/
[kN/t] (M) = 4.5
0,8 (M) = 5.0
Fn = 3.0
(M) = 5.5
5 (M) = 6.0
0,6 (M) = 6.5
4
Fn = 1.4
 3
0,4 Fn = 1.2 
2
Fn = 1.0
0,2 1
Fn = 0.8
0
4,5 5,0 5,5 6,0 6,5 7,0 0,5 1,0 1,5 2,0 2,5 3,0
(M) Fn

Fig. 3 RR/ = f(Fn,(M)) for L/B = 3.33 Fig. 4  = f(Fn,(M)) for L/B = 3.33

1,0

actual
0,8 Fn = 3.0
speed-dependent
RR/ + speed-independent
[kN/t]
0,6 (M) = 5.0
4 (M) = 5.5
(M) = 6.0
(M) = 6.5
0,4 3
Fn = 1.4 (M) = 7.0

Fn = 1.2  2
0,2
Fn = 1.0 1
Fn = 0.8
0,0 0
5,0 5,5 6,0 6,5 7,0 7,5 0,5 1,0 1,5 2,0 2,5 3,0
(M) Fn

Fig. 5 RR/ = f(Fn,(M)) for L/B = 5.41 Fig. 6  = f(Fn,(M)) for L/B = 5.41
 actual
speed-dependent
0,90 + speed-independent
(M) = 6.5
0,75 (M) = 7.0
RR/ (M) = 7.5
[kN/t] (M) = 8.0
0,60
Fn = 3.0 (M) = 8.5
2

0,45 


0,30 1
Fn = 1.4
Fn = 1.2
0,15
Fn = 1.0
Fn = 0.8
0
6,5 7,0 7,5 8,0 8,5 0,5 1,0 1,5 2,0 2,5 3,0
(M) Fn 

Fig. 7 RR/ = f(Fn,(M)) for L/B = 7.50 Fig. 8  = f(Fn,(M)) for L/B = 7.50

Towing power Dynamic trim Wetted surface

L/B 3.33 4.54 5.41 6.25 7.50 L/B 3.33 4.54 5.41 6.25 7.50 coefficient
Number Number Number
of errors s-i 1 9 33 6 5 of errors s-i - 1 10 3 - of errors
between between between
3 to 5 % s-d 1 10 32 4 5 0.2 to 0.3 s-d - 1 10 4 - 0.2 to 0.3 % 13
Number Number Number
of errors s-i - 1 2 - 1 of errors s-i - - 1 1 - of errors
above above above
5% s-d - 1 2 2 1 0.3 s-d - - 1 1 - 0.3 % 6
Number of Number of Number of
observations 123 152 119 164 120 observations 108 132 108 156 108 observations 61

Table 6 Discrepancies between calculated data and data given in Reference [1]
(s-i - speed-independent s-d - speed-dependent)
 APPENDIX 4 – RESISTANCE AND TRIM OUTPUTS

speed-dependent speed-dependent
+ speed-independent + speed-independent

0,20 (M) = 6.5 0,20

(M) = 6.0
(M) = 5.5
(M) = 6.5
(M) = 5.0
0,15 0,15 (M) = 6.0
(M) = 4.5
(M) = 5.5
RT/ RT/ (M) = 5.0

0,10 0,10

0,05 0,05

0,00 0,00
1,0 1,5 2,0 2,5 3,0 1,0 1,5 2,0 2,5 3,0
Fn Fn

Fig. 9 (RT /)100000=f(Fn,(M)) for L/B = 4Fig. 10 (RT /)100000=f(Fn,(M)) for L/B = 5

speed-dependent
speed-dependent
+ speed-independent
+ speed-independent

4
5 (M ) = 4.5
(M ) = 5.0
(M ) = 5.0 (M ) = 5.5
(M ) = 5.5 (M ) = 6.0
3 (M ) = 6.5
4 (M ) = 6.0 
 (M ) = 6.5 

3
2

1
1

0 0
1,0 1,5 2,0 2,5 3,0 1,0 1,5 2,0 2,5 3,0
Fn Fn

Fig. 11  = f(Fn,(M)) for L/B = 4 Fig. 12  = f(Fn,(M)) for L/B = 5

View publication stats