No - Ntnu Inspera 54166542 26312328

Tone Dale 2020 Master's thesis
NTNU
Norwegian University of
Science and Technology
Faculty of Engineering
Master's thesis
Department of Marine Technology
July 2020
Tone Dale
Performance Calculations
Development of Simplified
Methods for Ship Powering
Development of Simplified Methods for
Ship Powering Performance Calculations
Tone Dale
Marine Technology
Submission date: July 2020
Supervisor: Sverre Steen
Co-supervisor: Helene Muri
Norwegian University of Science and Technology

NTNU Trondheim
MASTER THESIS IN MARINE TECHNOLOGY
SPRING 2020
FOR
Tone Dale
Development of simplified methods for ship powering performance

calculations
As part of the KPN research project CLIMMS, NTNU is developing a computational model for the fuel
consumption and emissions to air from the world shipping fleet (or subsets of it). The model is using AIS
data, combined with a database with ship information (main dimensions, main engine size etc.). Since
there is an enormous number of ships, CFD and other complicated models for computing the required
propulsion power for the observed speed of each individual ship cannot be used, and experience shows
that the commonly used simple empirical models like Hollenbach and Holtrop aren’t suitable for all
relevant ship types and sizes. Also, the ship data in the database might not contain all the required data for
the existing empirical methods for power prediction. However, there are large amounts of in-service data
available that might be used to develop improved empirical models.
The project and master thesis will be performed in cooperation with the KPN project CLIMMs. The
project will contribute with knowledge and guidance, database access as well as data from ship operation.
The objective of the master thesis is establish and subsequently validate a ship powering performance
models suitable for use in the mentioned global fleet emissions calculation. The developed method can be
composed of existing methods, or it might partly or wholly be newly developed.
In the thesis the candidate shall present her personal contribution to the resolution of problem within the
scope of the thesis work.
Theories and conclusions shall be based on mathematical derivations and/or logic reasoning identifying the
various steps in the deduction.
The thesis work shall be based on the current state of knowledge in the field of study. The current state of
knowledge shall be established through a thorough literature study, the results of this study shall be written
into the thesis. The candidate should utilize the existing possibilities for obtaining relevant literature.
The thesis shall be organized in a rational manner to give a clear exposition of results, assessments, and
conclusions. The text should be brief and to the point, with a clear language. Telegraphic language should
be avoided.
The thesis shall contain the following elements: A text defining the scope, preface, list of contents, summary,
main body of thesis, conclusions with recommendations for further work, list of symbols and acronyms,
reference and (optional) appendices. All figures, tables and equations shall be numerated.
NTNU Trondheim
The supervisor may require that the candidate, in an early stage of the work, present a written plan for the
completion of the work. The plan shall include a budget for the use of laboratory or other resources that will
be charged to the department. Overruns shall be reported to the supervisor.
The original contribution of the candidate and material taken from other sources shall be clearly defined.
Work from other sources shall be properly referenced using an acknowledged referencing system.
The thesis shall be submitted electronically (pdf) in Inspera:

- Signed by the candidate
- The text defining the scope (this text) (signed by the supervisor) included
Supervisor : Professor Sverre Steen

Advisors: : Helene Muri
Start : 09.03.2020
Deadline : 31.08.2020
Trondheim, 13.01.2020
Sverre Steen
Supervisor
Preface
This master thesis presents a study of simplified methods for ship powering calculations for the world fleet.
The work was completed during the spring/summer of 2020, and serve as the final contribution to my Master
of Science degree in Marine Technology.
I would like to thank my supervisor Sverre Steen for giving valuable feedback, and for taking the time to have
meaningful discussions throughout the process. Further, I would like to thank my advisor Helene Muri for
providing insight to your work at IndEcol and for facilitating access to the MariTEAM model. I am grateful to
take part in the research project CLIMMS. Finally, I would like to thank both Prateek Gupta and YoungRong
Kim for assisting me in the model development.
Trondheim, July, 2020
Tone Dale
Abstract
International shipping is the main contributor to world trade by carrying more than 80 % of transported goods.
Though emissions per unit of goods transported are lower than any other mode of transport, the fleet is expe-
riencing increasing pressure to reduce global emissions. NTNU has developed a maritime transport emission
assessment model (MariTEAM), as part of the research project CLIMMS - Climate mitigation in the maritime
sector. MariTEAM is a computational model for the fuel consumption and emissions to air from the world
shipping fleet. This thesis contributes by developing and validating a new ship powering performance method
suitable for the MariTEAM model.
The current state of knowledge in the field of global fleet-wide power predictions has been reviewed. In line
with the literature, various empirical methods are implemented in the model developed in this thesis, which
is subdivided into five modules. First, missing input parameters are estimated for the ship. The calm water
resistance is calculated, followed by added resistance in wind and waves. Then the propulsive efficiency is
determined before the final power is obtained. As a result, the new method can predict the propulsion power
in realistic sea-states, for a wide range of ships in the fleet, while requiring few input parameters.
A case study of seven diverse vessels is applied to validate the powering performance of the new model.
The validation data include model test reports, sea trial reports and in-service data from voyages. Based
on the study, a final new power prediction method is presented. Validation against model test reports and
sea trials show that the new model achieves powering predictions with a mean deviation of ± 3 % and
standard deviation of 6% for exact input parameters. When parameter estimates are applied for missing
input, deviations are within 10% for the power predictions. Validation against in-service data demonstrates
that the powering predictions and the in-service measurements correspond well if the correct loading condition
is applied.
Sammendrag
Internasjonal skipsfart er den største bidragsyteren til verdenshandelen, og frakter mer enn 80 % av trans-
portert gods. Selv om utslippene per enhet som transporteres er lavere enn for noen annen transport, op-
plever skipsfarten et økende press for å redusere de globale utslippene. Som en del av KPN-forskningsprosjektet
CLIMMS, har NTNU utviklet MariTEAM-modellen, en beregningsmodell for drivstofforbruk og utslipp fra ver-
dens skipsflåte. Denne avhandlingen bidrar med å utvikle og validere en ny ytelsesmetode for skip som er
egnet for MariTEAM-modellen.
Den nåværende kunnskapen om prediksjon av effektforbruk for skip er gjennomgått. I samsvar med litter-
aturen er ulike empiriske metoder implementert i modellen, som består av fem overordnede moduler. Først
estimeres manglende tekniske parametere for skipet. Deretter beregnes stillevannsmotstanden, etterfulgt
av motstandsøkning i vind og bølger. Så bestemmes propulsjonsvirkningsgraden, før det endelige effektfor-
bruket beregnes. Dermed kan den nye metoden estimere effektforbruket i realistiske sjøtilstander og for et
bredt spekter av skip i flåten, kun ved hjelp av få parametere.
En casestudie av syv ulike skip er benyttet for å validere prestasjonen til den nye metoden. Validerings-
grunnlaget inkluderer modelltest rapporter, prøvetursdata og in-service data for seilende skip. Valideringen
mot stillevannsberegninger viser at den nye metoden kan predikere effektforbruket med gjennomsnittlig avvik
på ± 3 % og med et standardavvik på 6%, hvis eksakte parametere er kjent. Når parameterestimatene an-
vendes for manglende parametere avviker effektberegningene opp til 10%. Validering av in-service data viser
at det målte effektforbruket og det estimerte effektforbruket samsvarer godt hvis korrekt lastkondisjon er brukt.
Table of Contents
Preface 1
Abstract 3
Sammendrag 5
Table of Contents 9
List of Tables 13
List of Figures 16
List of Equations 18
Symbols 19
1 Introduction 1
1.1 Background and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Outline of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Theory of Ship Powering 5

2.1 Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Calm Water Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Added Resistance due to Wind and Waves . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.1 Propulsive Efficiency in Ideal Conditions . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.2 Load Variation Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.3 Total Delivered Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Ship Powering Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Empirical Prediction Methods 15

3.1 Empirical Resistance Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.1 Guldhammer-Harvald . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.2 Holtrop-Mennen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.3 Hollenbach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2 Added Resistance due to Wind and Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2.1 STAWAVE-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.2 STAWAVE-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2.3 STA-JIP wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.4 Fujiwara . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.5 Blendermann . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2.6 Townsin & Kwon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Propulsion Prediction Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3.1 Propulsive Efficiencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3.2 Empirical Formulas for the Propulsive Efficiency . . . . . . . . . . . . . . . . . . . . . 28
7
3.3.3 Prediction Based on Sea-Margin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.4 Load Variation Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.4 Complete Power Prediction Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.4.1 ShipCLEAN by Tillig et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.4.2 A Power Prediction model by Kristensen et al. . . . . . . . . . . . . . . . . . . . . . . 31
3.4.3 STEAM 3 by Jalkanen et al. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4.4 A Power Prediction model by Lindstad et al. . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Methods to Estimate Input Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5.1 Hull Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5.2 Propulsion Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4 The MariTEAM Model 35

4.1 Data Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.1 Sea-web . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.1.2 Additional Databases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1.3 Methods to Fill In Missing Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1.4 AIS Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.1.5 Weather Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 Power Prediction Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2.2 Added Resistance in Wind and Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2.3 Propulsive Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2.4 Total Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.5 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5 New Performance Prediction Method 43

5.1 Selection of Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.1.2 Added Resistance in Wind and Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.1.3 Propulsion Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.1.4 Input Parameter Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.1.5 Potential Improvements to the Current MariTEAM model . . . . . . . . . . . . . . . . 52
5.1.6 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.2 Program Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.2.1 Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.2.2 Flowchart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2.3 Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3 Verification of Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6 Results and Validation 59

6.1 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.2 Case Vessel Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3 Weather Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.4 Validation Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.5 Validation of Calm-Water Power Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.5.1 Case 1: Cargo Ship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.5.2 Case 2: Container Ship (13,000 TEU) . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.5.3 Case 3: Vehicles Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.5.4 Case 4: Wellboat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.5.5 Case 5: Chemical Tanker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.5.6 Case 6: Container ship (3,500 TEU) . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.5.7 Case 7: Bulk Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.6 Validation of In-service Power Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.6.1 Case 1: Cargo Ship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.6.2 Case 2: Container Ship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.7 Summary of Results and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.7.1 Performance of the Calm Water Resistance Methods . . . . . . . . . . . . . . . . . . 85
6.7.2 Performance of the In-service Prediction . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.7.3 Performance of the Parameter Estimates . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.8 Comparing the New Model to the Current MariTEAM Model . . . . . . . . . . . . . . . . . . . 87
6.8.1 Case 1: Cargo Ship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.8.3 Case 3: Vehicles Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.8.4 Case 4: Wellboat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.8.5 Case 5: Chemical Tanker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.8.7 Case 7: Bulk Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.8.8 Summary of Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7 Discussion 99
7.1 The New Performance Prediction Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.2 Validation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.2.1 Calm Water Powering Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.2.2 In-service Powering Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.2.3 Comparing the New and Current MariTEAM model . . . . . . . . . . . . . . . . . . . 102
8 Conclusion 103
9 Recommendations for Further Work 105
Bibliography 105
Appendix A Hollenbach Coefficients I
Appendix B Guldhammer-Harvald Coefficients III
Appendix C Holtrop-Mennen Coefficients V
Appendix D Sea-web Parameters VII
Appendix E Blendermann Wind Coefficients XI

E.1 Car Carrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI
E.2 Container Vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII
E.3 Tanker (loaded) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII
E.4 Cargo Vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII
E.5 Passenger Ship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII
E.6 Offshore Supply Vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIV
E.7 Ro-Ro/Lo-Lo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIV
E.8 Deep Sea Drilling Vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV
E.9 Research Vessel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV
Appendix F The New Program (MATLAB code) XVII

List of Tables
3.1 Empirical resistance procedures with limited applicable area . . . . . . . . . . . . . . . . . . . 15

3.2 Required and optional input parameters for the Guldhammer-Harvald method . . . . . . . . . 16
3.3 Recommended range for speed and main dimensions in Guldhammer-Harvald . . . . . . . . 16
3.4 Required and optional input parameters for Holtrop-Mennen’s method . . . . . . . . . . . . . 17
3.5 Recommended range for speed and main dimensions in Holtrop-Mennen . . . . . . . . . . . 18
3.6 Required and optional input parameters for Hollenbach’s method . . . . . . . . . . . . . . . . 19
3.7 Recommended range for main dimensions in Hollenbach (*) Extended range for mean resis-
tance calculation by Birk (2019) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.8 Required input parameters for STAWAVE-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.9 Required input parameters for STAWAVE-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.10 Permissible range for STAWAVE-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.11 Required input parameters for STAJIP wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.12 Required input parameters for Fujiwara wind resistance coefficients . . . . . . . . . . . . . . . 23
3.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.14 Required input parameters for the Kwon method . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.15 Propeller series with limited applicable area . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.16 Empirical Propulsive Efficiency Prediction formulas . . . . . . . . . . . . . . . . . . . . . . . . 28
3.17 Estimation formulas for hull parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.18 Estimation formulas for superstructure dimensions . . . . . . . . . . . . . . . . . . . . . . . . 34
3.19 Estimation formulas for propeller dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1 Relevant Sea-web parameters for the power performance predictions . . . . . . . . . . . . . . 37

4.2 ECMWF Weather Data Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.1 Number of ships in the fleet [%] passing the requirement of each method, all three methods,
one of the three methods, or either Hollenbach or Holtrop-Mennen. . . . . . . . . . . . . . . . 43
5.2 Air resistance coefficient values for container ships, tankers and bulk carriers as recommended
by Kristensen et al. (2017) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.3 Selected methods to predict the increased resistance in wind and waves . . . . . . . . . . . . 48
5.4 Comparison of the applicable area for the wind resistance methods . . . . . . . . . . . . . . . 49
5.5 Estimation formulas for superstructure dimensions . . . . . . . . . . . . . . . . . . . . . . . . 51
5.6 Estimation formulas for hull dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.7 Estimation formulas for propeller dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.8 Potential Improvements to the Current MariTEAM model . . . . . . . . . . . . . . . . . . . . . 52
6.1 Overview of case vessel data in the case study . . . . . . . . . . . . . . . . . . . . . . . . . . 59

6.2 In-service weather data variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.3 Main particulars of Case 1: Cargo ship at design draught . . . . . . . . . . . . . . . . . . . . 61
6.4 Deviation between calculated power and model test power for Case 1 (exact input parameters) 62
6.5 Deviation of estimated parameters for Case 1 with design loading condition . . . . . . . . . . 63
6.6 Main particulars of Case 2: Container Ship (13,000 TEU) at ballast draught . . . . . . . . . . 64
6.7 Deviation between calculated power and sea trial power for Case 2 (exact input parameters) . 65
6.8 Deviation of estimated parameters for Case 2 with heavy ballast loading condition . . . . . . . 66
6.9 Main particulars of Case 3: Vehicles carrier at design draught . . . . . . . . . . . . . . . . . . 67
11
6.11 Deviation of estimated parameters for Case 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.12 Main particulars of Case 4: Wellboat at design draught . . . . . . . . . . . . . . . . . . . . . . 69
6.15 Main particulars of Case 5: Chemical tanker at design draught . . . . . . . . . . . . . . . . . 72
6.18 Main particulars of Case 6: Container ship (3,500 TEU) at design draught . . . . . . . . . . . 74
6.21 Main particulars of Case 7: Bulk Carrier at ballast draught . . . . . . . . . . . . . . . . . . . . 77
6.22 Deviation between calculated power and sea trial power for Case 7 (exact input parameters) . 78
6.24 In-service weather data range for Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.25 RMSE for combinations of added resistance methods applied to Case 1 with in-service loading
condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.26 Deviation of predicted power for in-service loading condition and for maximum loading condition 82
6.27 In-service weather data range for Case 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.28 RMSE for combinations of added resistance methods applied to Case 2 with in-service loading
condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.29 Deviation of predicted power for in-service loading condition and for maximum loading condition 85
6.30 Applicable methods for Case 1 - Case 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.31 Mean deviation and mean std. deviation for the calculated resistance, propulsive efficiency and
power. Results include all cases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
power. Results only include cases where the respective methods are valid. . . . . . . . . . . . 86
6.33 Mean deviation and mean standard deviation of the estimated parameters . . . . . . . . . . . 87
power. Results include all cases with estimated parameters. . . . . . . . . . . . . . . . . . . . 87
6.35 Deviation of power predicted by the model test, new model, and the current MariTEAM model
for Case 1 (parameter estimates included). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.36 Deviation of power measured in-service, predicted by the new model, and the current MariTEAM
model for Case 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.37 Deviation of power predicted by the sea trial, new model, and the current MariTEAM model for
Case 2 (parameter estimates included). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.38 Deviation of calm water resistance predicted by the sea trial, new model, and the current
MariTEAM model for Case 2 (parameter estimates included). . . . . . . . . . . . . . . . . . . 91
6.39 Deviation of power measured in-service, predicted by the new model, and the current MariTEAM
model for Case 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.44 Deviation of power predicted by the sea trial, new model, and the current MariTEAM model for
Case 7 (parameter estimates included). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.45 Deviations for the calculated resistance, propulsive efficiency and power. Results include all
cases with estimated parameters, for the new model and the current MariTEAM model. . . . . 97
7.1 Mean deviation and mean std. deviation for the calculated power (Results are computed with
exact input parameters). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
E.1 Car carrier wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI

E.2 Container vessel wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII
E.3 Tanker (loaded) wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII
E.4 Cargo vessel wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII
E.5 Passenger ship wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII
E.6 Offshore supply vessel wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIV
E.7 Ro-Ro/Lo-Lo wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIV
E.8 Deep sea drilling vessel wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV
E.9 Research vessel wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV
List of Figures
1.1 Fleet wide CO2 emissions in 2015 from Olmer et al. (2017), International Council on Clean
Transport (ICCT, 2017). Data from exactEarth, IHS and ArcGIS. . . . . . . . . . . . . . . . . 1
2.1 Flow features of a moving ship based on Molland et al. (2017) . . . . . . . . . . . . . . . . . . 5

2.2 Decomposition of ship resistance components based on Molland et al. (2017) . . . . . . . . . 6
2.3 Typical wavelength dependence of added resistance Rwave in regular head sea waves. ⇣a is
the wave amplitude, B is the beam of the ship. (Faltinsen et al., 1980) . . . . . . . . . . . . . 10
2.4 Example of ISSC-spectrum with H1/3 = 8m and T = 10s, from Faltinsen (1993). (The energy
is distributed into the energy of N = 10 regular wave components) . . . . . . . . . . . . . . . 11
3.1 Definition of LBW L in STAWAVE-1 (ISO Technical Committee, 2015) . . . . . . . . . . . . . . 20

3.2 Definition of structural parameters in Fujiwara et al. (2017) . . . . . . . . . . . . . . . . . . . . 23
3.3 Applicable ranges for Emersons formula, inclduing the extension by Watson (1998) . . . . . . 29
4.1 Modules in the MariTEAM model - From ship to fleet by Bouman et al. (2016) . . . . . . . . . 35
4.2 Flowchart of the MariTEAM power prediction model . . . . . . . . . . . . . . . . . . . . . . . 36
4.3 Distribution of ship types in the merchant fleet by IHS (2020) . . . . . . . . . . . . . . . . . . 37
4.4 Scatter plot of ldt against potential predictor parameters by Ringvold (2017) . . . . . . . . . . 39
4.5 The MariTEAM AIS data track completer (Containerships, 2017). Figure courtesy of Radek
Lonka, IndEcol, NTNU. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.6 Range of validity for minimum resistance in Hollenbach’s method by Schneekluth and Bertram
(1998) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1 Comparison of mean and standard deviation of the total resistance predicted by Holtrop-
Mennen, Guldhammer-Harvald and Hollenbach by Hollenbach (1997), retrieved from Steen
et al. (2016) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2 Comparison of the STAWAVE methods and irregular wave model tests for a 174 m tanker
(MARIN, 2006) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 Definition of LBW L in STAWAVE-1 (ISO Technical Committee, 2015) . . . . . . . . . . . . . . 49
5.4 Conceptual flowchart of the new program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.5 Flowchart of the new power prediction method . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.6 Relative wave angle in the ship reference frame . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.7 Relative wind angle in the ship reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.1 In-service routes for the recorded voyage data . . . . . . . . . . . . . . . . . . . . . . . . . . 60

6.2 Validation of power prediction for Case 1 with design loading and exact input parameters.
Model test results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.3 Validation of power prediction for Case 1 with design loading and estimated parameters. Model
test results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.4 Validation of power prediction for Case 2 with heavy ballast loading and exact parameters. Sea
trial results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.5 Validation of power prediction for Case 2 with heavy ballast loading and estimated parameters.
Sea trial results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.6 Validation of power prediction for Case 2 with heavy ballast loading and estimated parameters,
with corrected propeller diameter. Sea trial results and calculated results. . . . . . . . . . . . 66
15
6.7 Validation of power prediction for Case 3 with design loading and exact parameters. Model test
results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.15 Validation of power prediction for Case 7 with heavy ballast loading and exact parameters. Sea
trial results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.16 Validation of power prediction for Case 7 with heavy ballast loading and estimated parameters.
Sea trial results and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.17 Validation of power prediction for Case 7 with heavy ballast loading, estimated parameters but
corrected propeller diameter. Sea trial results and calculated results. . . . . . . . . . . . . . . 79
6.18 Validation of power prediction for Case 1 with in-service loading. In-service measurements and
calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.19 The added resistance modules with highest accuracy (a) and lowest accuracy (b) in terms of
RMSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.20 Validation of power prediction for Case 1 with maximum loading from Sea-web. In-service
measurements and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.21 Validation of power prediction for Case 2 with in-service loading. In-service measurements and
calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.22 The added resistance modules with highest accuracy (a) and lowest accuracy (b) in terms of
RMSE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.23 Validation of power prediction for Case 2 with maximum loading from sea-web. In-service
measurements and calculated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.24 Comparison of power predicted by the model test, new model, and the current MariTEAM
model for Case 1 (parameter estimates included). . . . . . . . . . . . . . . . . . . . . . . . . 88
6.25 Comparison of measured in-service power to the power predicted by the new model (a) and
the current MariTEAM model (b) for Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.26 Comparison of power predicted by the sea trial, new model, and the current MariTEAM model
6.27 Comparison of calm water resistance predicted by the sea trial, new model, and the current
MariTEAM model for Case 2 (parameter estimates included). . . . . . . . . . . . . . . . . . . 91
6.28 Comparison of measured in-service power to the power predicted by the new model (a) and
the current MariTEAM model (b) for Case 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.33 Comparison of power predicted by the sea trial, new model, and the current MariTEAM model
A.1 Coefficients for computing CR in Hollenbach as presented by Birk (2019) . . . . . . . . . . . . I

List of Equations
2.1 Total calm water resistance coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Resistance coefficient components, MARINTEK . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Resistance coefficient components, ITTC’78 . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Viscous Resistance Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.5 Frictional resistance coefficient (ITTC’57) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.6 MARINTEK roughness correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.7 Townsin roughness correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.8 Base drag coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.9 Air resistance coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.10 Total resistance under the representative sea condition . . . . . . . . . . . . . . . . . . . . . 10
2.11 Total resistance in a seaway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.12 Wind resistance formula by ITTC (2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.13 Propulsive efficiency in ideal conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.14 Hull efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.17 Corrected propulsive efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.18 Total delivered brake power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1 Guldhammer-Harvald total resistance formula . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 Holtrop-Mennen total resistance coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 Hollenbach total resistance coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.7 Added resistance in waves by STAWAVE-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.8 STAWAVE-2 Added wave resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.12 Added resistance due to wind by ISO15016 . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.13 Weather Speed Adjustment by Townsin and Kwon (1983); Kwon (2008) . . . . . . . . . . . . 24
3.16 Thrust loading coefficient by Breslin and Andersen (1994) . . . . . . . . . . . . . . . . . . . . 26
3.20 Wake and thrust deduction factor by Harvald (1992) . . . . . . . . . . . . . . . . . . . . . . . 27
3.21 Holtrop (1977) thrust deduction factor estimate . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.22 Holtrop (1977) wake factor estimate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.23 Hollenbach model scale hull efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.24 Hollenbach model scale wake fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.25 Hollenbach full scale wake fraction, according to ITTC’78 . . . . . . . . . . . . . . . . . . . . 28
3.26 Relative rotative efficiency by Holtrop (1977) . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.27 Approximated power available in ideal conditions . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.28 Approximated propulsive efficiency in ideal conditions . . . . . . . . . . . . . . . . . . . . . . 29
3.31 Modified thrust loading coefficient by Breslin and Andersen (1994) . . . . . . . . . . . . . . . 30
3.32 Modified propulsion efficiency factor by Lindstad et al. (2014) . . . . . . . . . . . . . . . . . . 30
4.1 Estimation of total resistance in the MariTEAM model . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Frictional resistance (ITTC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Frictional resistance coefficient (ITTC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4 Estimated wet surface area by the Mumford formula . . . . . . . . . . . . . . . . . . . . . . . 41
4.5 Hollenbach residual resistance coefficient (minimum resistance) . . . . . . . . . . . . . . . . 41
4.6 Weather Speed Adjustment by Townsin and Kwon (1983); Kwon (2008) . . . . . . . . . . . . 41
4.7 Propulsive efficiency factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.1 Total resistance decomposition in the new method . . . . . . . . . . . . . . . . . . . . . . . . 45
5.5 MARINTEK form factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.18 Guldhammer-Harvald residuary resistance coefficient with corrections . . . . . . . . . . . . . 47
5.21 Guldhammer-Harvald bulb correction formula for tankers/bulk carriers by Kristensen et al. (2017) 47
17
5.22 Guldhammer-Harvald bulb correction formula for container ships by Kristensen et al. (2017) . 47
Symbols
Hull and Propeller Characteristics

Lwl - Length of waterline
Lpp - Length between perpendiculars
Los - Length of surface
Loa - Length over all
B - Ship breadth
T - Ship draught
r - Volume displacement
ldt - Light displacement tonnage
dwt - Deadweight
S - Wetted surface area
SB - Wetted base/transom area
Sapp - Wetted surface of appendages
AV T - Transverse projected area above the waterline
AV L - Longitudinal projected area above the waterline
D - Depth from keel to uppermost continuous deck
ABT - Transverse cross section area of bulb
CB - Block coefficient
CM - Midship section coefficient
CP - Prismatic coefficient
Cstern - Stern shape parameter in Holtrop-Mennen
LCB - Londitudinal center of buoyancy
VD - Design speed
VW - Reduced speed corrected for added resistance
Dp - Propeller diameter
AE /A0 - Propeller expanded ratio
n - Propeller rpm
Resistance
CT - Total resistance coefficient
CR - Residuary resistance coefficient
k - Form factor
CF - Frictional resistance coefficient
CF - Hull roughness correction
CDB - Base drag coefficient
CAA - Air resistance coefficient
CApp - Appendage resistance coefficient
CA - Correlation allowance
V - Ship speed
⇢s - Sea water density
⌫ - Sea water kinematic viscosity
Rtot - Total resistance in wind and waves
Rwind - Added resistance due to wind
Rwave - Added resistance due to waves
RAWML - Resistance due to induced wave motions
RAWRL - Resistance due to wave reflection
Fn - Froude number
h - Water depth
g - Constant of gravity
H1/3 - Significant wave height
⇣A - Wave amplitude
Propulsion
⌘D - Propulsion efficiency in ideal conditions
⌘tot - Propulsive efficiency in trial conditions
⌘0 - Open-water (propeller) efficiency
⌘H - Hull efficiency
⌘R - Relative rotative efficiency
⌘M - Mechanical losses
⌘S - Losses in the transmission system
PD - Delivered power at the propeller
t - Thrust deduction factor
w - Wake fraction
VA - Speed of inflow to the propeller
T - Torque
PS - Calm water power requirement
Ptot - Total power requirement
⇠P - Slope of the linear curve in the load variation test
Abbreviations
GT = Gross tonnage
IMO = International Maritime Organisation
ITTC = International Towing Tank Conference
MCR = Maximum continuous rating
SM = Sea margin
Chapter 1
Introduction
1.1 Background and Motivation

Shipping is the main contributor to world trade by carrying more than 80 % of transported goods. As of 2018,
the world merchant fleet counts about 94,000 vessels of 100 GT and above (UNCTAD, 2018). The main
part of the fleet is cargo-carrying ships (dry bulk, tankers, general cargo, container, passenger), followed by
non-cargo ships (fishing, research, offshore, and other). The largest ship segments transport cargo over
long ranges within a well-defined system of global shipping routes. Figure 1.1 illustrates the fleet-wide CO2
emissions from 2015, related directly to the vessel traffic densities on the sea routes.
Figure 1.1: Fleet wide CO2 emissions in 2015 from Olmer et al. (2017), International Council on Clean Transport (ICCT,
2017). Data from exactEarth, IHS and ArcGIS.
Merchant ships in international traffic are subject to regulations by the International Maritime Organization of
the United Nations (IMO). Despite ships having lower emission intensities per unit mass transported than any
other mode of transport (Edenhofer, 2014), it is recognized that the fleet represents a significant reduction
potential to contribute to the global efforts of limiting global warming (Hoegh-Guldberg et al., 2019). Although
international shipping is excluded from the Paris Agreement, IMO is pursuing the development and imple-
mentation of measures to address greenhouse gas (GHG) emissions from the fleet (Olmer et al., 2017). In
2018 they decided that maritime GHG emissions shall be reduced by 50 % (compared to 2008) within 2050.
Shipping is experiencing an increasing pressure to decarbonize and reduce emissions to air, on the way to
reach the 2-degree target of the Paris Agreement. As of 2020, new emission regulations are applied to the
world ship fleet as the limit for sulfur in fuel oil is reduced from 3.5 to 0.5 %.
In order to develop effective strategies for this green transition, IMO member states pursue understanding the
1
Chapter 1. Introduction
current trends in ship activity and emissions (Hoegh-Guldberg et al., 2019). Identifying the drivers of shipping
emissions is decisive in making informed decisions and influencing policymakers. The world shipping fleet
emissions can be estimated in terms of fuel consumption, either by top-down or bottom-up (activity-based)
approaches. Top-down methods are based on the reported marine fuel sales statistics, and ship-type specific
results are not obtained. Bottom-up methods estimate the fuel consumption for each ship based on power
produced by the engines. Such power predictions require ship technical and operational data. Due to the
enormous number of ships in the fleet, a simplified power prediction method with sufficient accuracy is needed
to obtain estimates across the whole fleet and within sub-segments.
Both the third IMO GHG study (Smith et al., 2014) and the 2017 ICCT study on GHG emissions from global
shipping, represent bottom-up studies with minor differences in the methodologies. Characteristic for these
bottom-up methods is the application of highly simplified empirical methods for the ship powering calcula-
tions. Even though empirical bottom-up methods are becoming increasingly accurate with improving AIS data
coverage (Olmer et al., 2017), there is a trade-off between the simplified calculations and the accuracy of
the results. There exist well established empirical methods like Hollenbach (1998) and Holtrop and Mennen
(1982), commonly used for power predictions. The International Towing Tank Conference (ITTC, 2017) is
a recognised worldwide association of hydrodynamics research organisations that operate towing tanks or
similar model test laboratories. ITTC establishes recommended procedures for powering predictions, includ-
ing empirical methods. However, several of these methods require many input parameters and may not be
suitable for all ship types.
In order to modify the existing methods, there are mainly two approaches found in the literature. A simple
method with few parameters is tuned to modern ships, or a detailed method is combined with smart estimates
for the required input parameters. Kristensen et al. (2017) studied the simple method of Guldhammer and
Harvald (1974) and assessed the accuracy of some of the parameters when applied to various ship types.
The study demonstrates a tuning of a historical method to present-day ship segments. Jalkanen et al. (2012)
combines several empirical methods in an assessment model of ship traffic exhaust emissions, ’STEAM3’. In
this study, the power predictions are based on Hollenbach and demonstrate how many input parameters can
be estimated for a traditional method with a higher level of detail. A similar model is ’ShipCLEAN’ by Tillig
(2020), which combines existing empirical formulas with new developed procedures to predict the ship power
performance.
Muri et al. (2019a,b) and Bouman et al. (2016) present global fleetwide emission predictions in the maritime
transport emission assessment model (MariTEAM). The model is developed as part of the KPN research
project ’CLIMMS’ (climate mitigation in the maritime sector), which is an interdisciplinary study connected
to SFI Smart Maritime. As a bottom-up model, MariTEAM applies empirical ship power prediction methods
for the emission calculations. The model input comprises ship technical data and AIS data combined with
hindcast weather data.
1.2 Objective
The main objective of this thesis is to identify, develop, and validate a new ship powering performance method
suitable for the MariTEAM model. This includes establishing the current state of knowledge in the field of
global fleet-wide power predictions. In line with the literature, various approaches to modifying empirical
methods will be assessed. Further, it includes validating the new model with data from model tests, sea trials
or in-service measurements for a range of vessel types.
1.3 Outline of Thesis

Chapter 2 presents a review of ship powering theory, in order to establish the most important effects on ship
resistance and propulsion. Chapter 3 identifies existing empirical procedures, which is narrowed down to the
most relevant. Chapter 4 outlines the current power predictions in the MariTEAM model. In addition, relevant
ship technical databases are investigated. Chapter 5 presents the methodology applied in selecting the em-
pirical methods for the new power prediction model. The new method is outlined, and potential improvements
to the current MariTEAM model is emphasized. All the empirical methods included in the model are verified,
2
Chapter 1. Introduction 1.3 Outline of Thesis
to ensure that these are implemented correct mathematically. In Chapter 6, a case study is conducted for
seven vessels. The results are validated against model test reports, sea trial reports and in-service data. In
addition, the performance of the new model and the current MariTEAM model is compared. Chapter 7 dis-
cusses the main findings and the performance of the new method. Chapter 8 presents the final conclusions,
before recommendations for further work are given in Chapter 9.
During the winter of 2019/2020, the project thesis was conducted as a preliminary study of candidate methods
for the ship powering calculations in the MariTEAM model. The study comprised a comprehensive literature
review of suitable methods and served as a basis for the current work. However, all relevant findings are
included and further elaborated on in this master thesis, which represents a complete and independent as-
sessment.
3 3
Chapter 1. Introduction
4
Chapter 2
Theory of Ship Powering
This chapter outlines the basic theory of ship powering based on ship resistance and total propulsive effi-
ciency. In order to develop a suitable power prediction model, the ship resistance and propulsive efficiency
should be determined with the highest possible accuracy. The objective of this chapter is to establish the
aspects of ship resistance and propulsive efficiency with significance for ship powering.
2.1 Resistance
As of today, there is a consensus for defining ship resistance in the context of ship hydrodynamics for a calm
water sea state. Ships are traditionally optimized for operation in this sea state, although ships travel most
of the time in wind, waves, and current. The ideal calm water condition neglects these effects, which is a
significant simplification of the real conditions in a seaway. According to Wartsila (2019), when navigating in
head-sea, the resistance can increase by 50-100% of the ship resistance in calm weather. In order to predict
the powering performance of ships in a seaway, the added resistance due to wind and waves, and the change
of propulsive efficiency must be taken into account. This is further elaborated on in Sections 2.1.2 and 2.2.2.
2.1.1 Calm Water Resistance

The calculation of calm water resistance for a moving ship can be based on two different approaches. Either
by considering energy dissipation from the hull or by evaluating the forces acting on the hull. The energy
dissipation can be observed in terms of the flow features developing around the hull, as presented in Figure
2.1 by Molland et al. (2017). There is a wave pattern moving with the hull and a wake of turbulent flow
extending behind the ship.
Figure 2.1: Flow features of a moving ship based on Molland et al. (2017)
Based on this method, the governing principle divides the total resistance into viscous resistance and wave-
making resistance from the wave pattern generated. It is further assumed that these are independent, which
is a practical simplification to illustrate the physical problem (Birk, 2019). Note that until 2017, ITTC applied
"Residual resistance" instead of "Wave making resistance" in their recommended procedure.
5
Chapter 2. Theory of Ship Powering
If the forces acting on the hull is considered, it is differed between pressure resistance, acting normal to the
hull and frictional resistance, acting as a shear force on the hull. The viscous ship resistance is defined as the
frictional resistance and viscous pressure resistance, corrected for the hull shape and fullness. Ship frictional
resistance is calculated as the equivalent resistance of a flat plate with the same Reynolds number, area,
and length, moving longitudinally through the water. Due to the ship volume, the velocity along the hull is
higher than the ship speed, which increases the viscous resistance relative to a plate (Steen et al., 2016).
The viscous pressure resistance is a pressure resistance due to viscous flow effects. It accounts for three-
dimensional flow effects such as flow separation due to appendages or in the aft hull shape. In accordance
with these definitions, the total resistance can be decomposed as illustrated by Molland et al. (2017) in Figure
2.2
Figure 2.2: Decomposition of ship resistance components based on Molland et al. (2017)
The resistance components are commonly expressed as dimensionless coefficients, as presented in Equation
2.1 for the total resistance coefficient.
RT
CT = (2.1)
0.5 · ⇢ · S · V 2
RT [N ] is the total resistance, ⇢ [kgm 3

] is the sea water density, S [m2 ] is the wetted area, and V is the ship
speed [ms 1 ].
MARINTEK restistance decomposition
Under the main assumption of resistance being divided into wave making (or residual) and viscous resistance,
there are several ways of decomposing the total resistance into smaller components. These include air
resistance, base drag, roughness effects, and more. In the current work, it is chosen to apply a decomposition
suggested by MARINTEK (2020), given in Equation 2.2.
CT = CR + (1 + k) (CF + CF ) + CDB + CAA + CApp + CA (2.2)
where the dimensionless coefficients are:
6
Chapter 2. Theory of Ship Powering 2.1 Resistance
CT - Total resistance
CR - Residuary resistance
k - Form factor
CF - Frictional resistance
CF - Hull roughness correction
CDB - Base drag
CAA - Air resistance
CApp - Appendage resistance
CA - Correlation allowance
MARINTEK applies a modification of the ITTC’78 procedure following Equation 2.3 (ITTC, 2017). In ITTC’78,
the base drag coefficient CDB is not included, and the hull roughness correction CF may include correlation
allowance. Further, the viscous resistance is defined differently in the two methods. As seen in Equation 2.2,
MARINTEK includes the roughness correction in the viscous resistance term.
CT = (1 + k)CF + CF + CA + CR + CAA (2.3)
The theory of the various ship resistance components will be briefly elaborated on in the following, based
on the basic division into components as outlined by MARINTEK. The respective calculation methods are
presented in Section 2.3.
Residuary resistance
Froude first defined the term residuary resistance in the 1860s as the remaining resistance when the friction
is subtracted from the total. Today the method is refined, and it is common to subtract all other non-Froude
scaled resistance components, in accordance with Equation 2.2. It is difficult to determine the residuary
resistance accurately, but the main contributions are from wave resistance and viscous pressure resistance.
The viscous pressure resistance represents the smaller contribution and is mainly due to flow separation
behind the hull. As presented previously, the ship generates a typical wave system that contributes to the
wave resistance. The interaction between the hull and the wave system is complex to evaluate but depends
strongly on the local shape (Schneekluth and Bertram, 1998). The wave resistance dominates the total
resistance for a fast, slender ship. According to Wartsila (2019), the residual resistance typically accounts for
the following amount of total resistance:
• 40-60% for high-speed ships (such as container ships and passenger ships)
• 10-25% for low-speed ships (such as bulk carriers and tankers)
It is therefore deemed important to predict the residual resistance with high accuracy.
Form factor
The calculation of frictional resistance assumes a flat plate, and the form factor is introduced to account for
the hull’s shape and fullness. Form factors commonly express the relation between the viscous resistance
CV and the frictional resistance, as presented in Equation 2.4. The value of the form factor can both refer to
k and (1 + k) (Steen et al., 2016). It is also important to note that the value of the form factor is related to the
friction line applied to find the frictional resistance CF .
CV = (1 + k) (CF + CF ) (2.4)
Several friction lines exist and the ITTC’57 correlation line is among the widely used methods. The frictional
coefficient is expressed empirically as a function of the dimensionless Reynolds number Rn , as presented in
Equation 2.5
0.075
CF = 2 (2.5)
(log(Rn ) 2)
7
A correction for form effect is included in the ITTC’57 formula and initially it was applied without an additional
form factor. However, today it is common practice to consider the ITTC’57 as a flat plate friction line and add
a form factor.
A number of empirical formulas to determine the form factor exist, including MARINTEK’s formula and Holtrop’s
formula. Since the formulas are determined by fitting a curve to scatter plots, results can range from upper
to lower estimates, i.e., include varying resistance contributions. According to Steen (2011) it is convenient to
apply a form factor which includes the viscous pressure if model tests are unavailable. The Holtrop formula
includes viscous pressure effects and is therefore considered relevant to evaluate in selecting the form factor
method. The selection of form factor for the power prediction model is presented in Section 5.1.
Frictional resistance
The frictional resistance of a full scale ship is computed as the resistance of a flat plate with the same speed,
area, and length, corrected for the increased frictional resistance due to hull roughness (CF + CF ). The hull
roughness is a function of coating type, fouling, fractures in the coating, and rust and damage from mechanical
devices. Fouling is the marine growth on the hull and can develop faster than other roughness contributions.
As fouling is a living organism, it depends on temperature, light, salinity, and a number of parameters, which
makes it challenging to make a reliable estimate of the increased resistance. The roughness is measured in
µm and increases over time. Typical roughness values are presented in the following. It is important to note
that the values are dependent on the frequency of docking and cleaning or recoating the hull.
Based on Steen et al. (2016):
• Newbuild vessel: 100 150µm

• Yearly growth rate due to rust and paint detoriation: 20µm
• 10-15 year old vessel: 300µm
Based on Townsin and Byrne (1980):
• Newbuild vessel: 80 120µm

• Yearly growth rate due to rust and paint detoriation:
10µm for high-performance coating and cathodic protection
75 150µm for resinous coatings and no cathodic protection
The total frictional resistance depends on the size of the wetted area of the ship, Reynolds number, and rough-
ness. However, above a certain Reynolds number, the roughness contribution dominates, and the frictional
resistance is no longer dependent on the Reynolds number (Steen et al., 2016). Wartsila (2019) estimate
the frictional resistance to represent 70 90% of the total resistance for low-speed ships and up to 40% for
high-speed ships.
According to Townsin and Kwon (1983), the deterioration from a good newbuilding hull surface (100µm) to a
typical in-service value (220µm) can result in the same added resistance as from wind and waves. Another
significant effect is the growth of roughness and fouling on the propeller, which can significantly reduce the
power performance of the ship further. It is therefore considered important to include the effects of roughness
in the fleetwide calculations. MARINTEK (2020) suggests the following roughness correction formula:
⇥ ⇤
CF = 110 · (H · V )0.21 403 · CF2 (2.6)
where H [µm] is the roughness. Other empirical formulas exist, such as Equation 2.7 according to Townsin
and Mosaad (1985), recommended by ITTC (2017).
"✓ ◆1/3 #
6
H · 10 1/3
CF = 0.044 10 · Rn + 0.000125 (2.7)
LW L
where LW L [m] is the ship waterline length. The roughness correction applied in the model is presented in
Section 5.1.
8
Base drag
Most ships have a partly submerged transom stern, which causes a separated flow at the transom. The
separation creates the base drag force, which can be represented dimensionless in accordance with Steen
et al. (2016) as Equation 2.8.
s
3
DB (SB /S)
CDB = 1 2 = 0.029 (2.8)
2 ⇢V S CF
where S [m] is the wetted surface area, excluding that of the transom, and SB is the base/transom/frontal
area. The formula is based on a body shaped like a projectile in infinite fluid and is valid for ships as long as
the speed is sufficiently low for the transom stern to be wetted. Base drag effects are important for ships with
low Froude numbers and large transoms.
Air resistance
The ship structure above the waterline (superstructure) is subject to air resistance, which depends on the
superstructure size, shape, and ship speed. The air resistance coefficient CAA can be calculated according
to the ITTC’78 procedure presented in Equation 2.9 by Birk (2019).
⇢air · AV T
CAA = CDA (2.9)
⇢·S
where CDA is the air drag coefficient, commonly determined by wind tunnel test data or by empirical esti-
mates. AV T is the transverse projected area above the waterline. Note that air resistance refers to a ship
traveling in still air, hence it does not account for wind. The resistance contribution is not significant for slow
ships.
Appendage resistance
Typical ship appendages like rudders, bilge keels, stabilizer fins, shaft brackets, and more, add to the ship
resistance. The resistance contribution is mainly frictional, although bluff or poorly aligned appendages may
cause flow separation. Molland et al. (2017) estimate the main appendages (rudder and bilge keel) resistance
to be 2-5% relative to the hull naked resistance of single-screw ships and it is predicted to be higher for twin
screw ships. Several empirical formulas to predict appendage resistance exist and these typically require
detailed characteristics of the appendages. Fortunately, most merchant ships only have a few appendages
and the difficulties in estimating effects of appendages are only significant for some unconventional ships
(Bertram, 2012). The error of neglecting this contribution is therefore considered relatively small.
Correlation factor
When resistance is calculated based on model tests, an empirical correlation factor CA is applied in the scaling
process. The factor depends on the model test and scaling method. It accounts for deviations between the
predicted resistance from the model test, and the calculated full-scale resistance from power measurements.
Some of the empirical resistance prediction methods are regression-based from model tank tests and include
correlation factors.
Shallow water correction
Correction of the ship performance can be made due to the effects of shallow water. Shallow water can
increase the frictional resistance and the wave resistance for the ship (Schneekluth and p Bertram, 1998). A
significant increase occurs at a depth h near the critical depth Froude number Fnh = V / gh = 1. It is difficult
to calculate the increase if the effect is strong, but simple corrections can be made for a weak influence e.g.
as suggested by Lackenby (1963).
9
2.1.2 Added Resistance due to Wind and Waves

The calculation of added resistance due to wind and waves is needed to predict the powering performance of
a ship in a seaway. Wind resistance is enforced on the superstructure, while wave resistance is generated by
wave-induced ship motions and wave reflection of the hull. The contributions add to the calm water resistance
from Equation 2.2 to predict the real conditions in a seaway. Equation 2.10 illustrates the summation as
suggested by ITTC (2018). In the current work, Rtot defines the total ship resistance, including weather
effects, and RT defines the total calm water resistance.
Rtot = RT + Rwind + Rwave (2.10)
Rtot - Total resistance in wind and waves

RT - Total calm water resistance
Rwind - Added resistance due to wind
Rwave - Added resistance due to waves
Added wave resistance
The added wave resistance can be further subdivided as Equation 2.11, composed of mean resistance due
to wave reflection RAWRL and mean resistance due to induced wave motions RAWML .
Rwave = RAWRL + RAWML (2.11)
The predictions of added resistance from a specific region or route depend on seastate and weather data.
Regarding ship characteristics, the added wave resistance is generally more dependent on ship size than the
ship shape (Schneekluth and Bertram, 1998). For ships with a large length L relative to the wavelength ,
wave reflection dominates the wave resistance. This is typically the case for ships in head waves if the sea
state is mild, and the wave height is restricted (ISO Technical Committee, 2015). If the ship’s length is short
relative to the wavelength, wave-induced motions are significant and must be considered. Faltinsen et al.
(1980) presents a relation between typical wavelengths and added resistance for ships in regular head sea
waves in Figure 2.3.
Figure 2.3: Typical wavelength dependence of added resistance Rwave in regular head sea waves. ⇣a is the wave
amplitude, B is the beam of the ship. (Faltinsen et al., 1980)
The illustrated relation considers regular waves, a simplified representation of the irregular sea that the ship
meets. In order to calculate the mean wave resistance in irregular seas, it is common to simulate the irregular
sea as a number of regular wave components and summarise the wave loads from each component by linear
theory. The sea state can be evaluated in terms of a wave energy spectrum S(w) as a function of circular
frequency w.
10
Figure 2.4: Example of ISSC-spectrum with H1/3 = 8m and T = 10s, from Faltinsen (1993). (The energy is distributed
into the energy of N = 10 regular wave components)
The modified Pierson-Moskowitz (ISSC) spectrum in Figure 2.4 by Faltinsen (1993) demonstrate a widely-
used model, recommended by the ITTC for fully-developed seas. It is characterised by two sea state pa-
rameters, the significant wave height H1/3 , and wave period T . Several empirical methods to predict wave
resistance exist and are further elaborated on in Section 3.2.
Added resistance in wind
The wind resistance enforced on the superstructure depends on the projected area above the waterline, wind
direction, and velocity. The resistance increase is significant for ships like container ships and ferries, with
large areas above the waterline. Wind resistance coefficients depend on ship type, shape, and geometry, as
well as the relative wind direction. Equation 2.12 presents the resistance increase according to ITTC (2014).
1 2
Rwind = ⇢A VWR CX ( WR ) AV T 0.5⇢A · CX (0) · AVT · VG2 (2.12)
2
⇢A - Mass density of air

VWR - Relative wind speed
CX - Wind resistance drag coefficient
W R - Relative wind direction
AV T - Transverse cross-sectional area above waterline
VG - Ship speed over ground
The wind resistance drag coefficients are derived from wind tunnel tests or may be determined by empirical
data sets. As values for wind velocity and direction vary with time, mean values over specific periods are
commonly applied in calculations. Calculation procedures are presented in Section 3.2.
Sea margin
In the design stages of a ship, it is common practice to account for the speed loss due to wind and waves by
using a sea margin of 15 % on the power (SFI, 2016). However, this margin often accounts for other service
condition effects as well (like roughness). Current standard procedures can calculate effects of wind to still air
conditions with confidence, however it is difficult to make corrections for waves with a high level of accuracy
(Townsin and Kwon, 1983). Despite this, there exist several computational and empirical methods, which will
be presented in Section 3.2.
11
2.2 Propulsion
The propulsive efficiency ⌘D must be determined in order to calculate the necessary machinery power of the
ship. It expresses the relationship between the effective power and the power developed by the propeller PD in
ideal conditions. Since the propulsion system and the ship hull interacts, the flow field changes. The propeller
inflow is influenced by the hull upstream, and the presence of the propulsion system itself changes the aft hull
flow. By taking the effect of these interactions into account, the propulsive efficiency can be determined in
accordance with Equation 2.13 (Schneekluth and Bertram, 1998).
RT · V
⌘D = ⌘0 · ⌘H · ⌘R = (2.13)
PD
⌘0 - Open-water (propeller) efficiency

⌘H - Hull efficiency
⌘R - Relative rotative efficiency
PD - Delivered power at the propeller
RT - Total calm water resistance
V - Ship speed
2.2.1 Propulsive Efficiency in Ideal Conditions

The propulsive efficiency coefficients are commonly evaluated in ideal calm water conditions. However, it is
well known that the propulsive efficiency changes in a seaway and the amount of change depends on the
added resistance. The following sections establish the efficiency components in ideal conditions, and the
effect of load variation in a seaway is outlined in Section 2.2.2.
Propeller efficiency
The propeller efficiency ⌘0 evaluates the propeller’s performance in open water, i.e., without the influence of
the hull. It is often determined by model tests but can be estimated empirically based on propeller series data
or by the use of different numerical methods. According to Steen et al. (2016), the propeller efficiency for
conventional propellers in design condition is typically in the range of 0.6-0.8.
Hull efficiency
The hull efficiency ⌘H is defined by Equation 2.14
1 t
⌘H = (2.14)
1 w
w is the effective mean wake fraction, accounting for the speed reduction from the ship speed V to the inflow
to the propeller VA . It is defined as
w=1 VA /V (2.15)
The thrust deduction factor t accounts for the increased resistance on the hull introduced by the working
propeller. The resistance is mainly due to the propeller accelerating the water inflow, and the increased speed
results in reduced propeller efficiency. The thrust factor is defined as
t=1 RT /T (2.16)
where RT is the total resistance and T is the propeller thrust. Both the thrust factor and the wake fraction are
often determined by model tests, but can be found by empirical methods. The hull efficiency for single screw
ships are usually larger than 1.0, and typically in the range of 1.05-1.1 (Birk, 2019).
Relative rotative efficiency
The relative rotative efficiency ⌘R accounts for the variable propeller blade loads due to the non-homogenous
wakefield inflow. It is normally in a narrow range of 0.97-1.03 (Steen et al., 2016).
12
Chapter 2. Theory of Ship Powering 2.3 Ship Powering Methods
2.2.2 Load Variation Correction

The propulsive efficiency changes due to load variations introduced by the added resistance in a seaway.
Increased load on the propeller due to wind and waves usually decreases the efficiency. According to Valanto
and Hong (2017), the propulsive efficiency losses are strongly connected to the decreasing open-water effi-
ciency ⌘0 , under increased propeller loading in waves. The propeller efficiency is reduced as a result of the
changed propulsion point. Methods to predict the change in the propeller efficiency are presented in Section
3.3.4.
The changes in hull interaction factors such as wake and thrust deduction are less known effects (SFI, 2016).
According to Tillig (2020), the changes in thrust factor, wake fraction, and relative rotative efficiency are small
and it may be reasonable to neglect these effects.
The final corrected propulsive efficiency is denoted as ⌘tot in the current work, and is defined by Equation
2.17.
(RT + Rwave + Rwind ) · V

⌘tot = ⌘0,corrected · ⌘H · ⌘R = (2.17)
PD
2.2.3 Total Delivered Power

When the total resistance and the propulsive efficiency is determined, the final power requirement can be
calculated. The total delivered brake power PB must overcome mechanical losses, losses in the transmission
system due to shafts and bearings, and is finally determined by Equation 2.18.
PD Rtot · V
PB = = (2.18)
⌘G · ⌘S ⌘G · ⌘S · ⌘tot
⌘G - Gear box losses

⌘S - Losses in the transmission system
By knowing the propulsive efficiency and losses in mechanical and transmission systems, and by calculating
the ship resistance at a certain speed V , the required power can be found. According to Schneekluth and
Bertram (1998), the shaft efficiency ⌘S is typically 0.98-0.985. If the system is fitted with gears, ⌘G is usually
larger or equal to 0.95 (Birk, 2019).
2.3 Ship Powering Methods

The ship powering evaluation methods range from traditional model tests to advanced numerical (CFD) meth-
ods. Generally, these methods are applied at a design stage in the absence of a prototype to test at full scale.
Model tests are well established and considered to be a reliable performance prediction method at a design
stage. Numerical analyses, i.e., CFD simulations, are of increasing significance, but are time-consuming and
not as reliable as model tests unless the user has significant experience. Therefore, model tests and CFD
simulations are not considered relevant to develop simplified prediction methods and are not further studied
in the current work.
In order to compute the required power for the observed speed of each ship in the world fleet, empirical
methods represent the most suitable approach. Historically, many systematic ship model tests and propeller
series have been conducted. The empirical regression methods available are results of regression analyses
of data from these model tests and full-scale ship trials. Most estimates are simple but have limited accuracy
and application area. Despite a limited accuracy of ±10%, the empirical regression methods are generally
the method of choice in early design stages (Birk, 2019).
13
Some of the old regression-based empirical methods can be re-visited and modified to apply to fleet-wide
calculations of ships in service. For some of the methods having general applicability, there is a considerable
variation in the level of detail, i.e., the required number of input parameters for the calculations. This applies to
both resistance estimates and propulsive efficiency predictions. A review of the relevant empirical prediction
procedures will be presented in the next chapter.
14
Chapter 3
Empirical Prediction Methods
This chapter presents the relevant existing ship powering performance prediction methods. As discussed in
Section 2.3 it is focused on computationally simple models with a wide application area, while requiring limited
input. Methods to predict resistance, propulsion and the change of performance due to wind and waves are
included. A comparison of these methods is presented in Section 5.1. Some parts of this chapter are based
on a comprehensive literature study from the project thesis (Dale, 2020), although outlined in a more concise
version in the following.
3.1 Empirical Resistance Procedures

A wide range of empirical resistance methods are identified, and systematic series and regression analysis
are amongst the main approaches to empirically predict ship power requirements. The relevance of the well-
known procedures is reviewed in the following. The empirical methods with limited applicable areas are listed
in Table 3.1.
Table 3.1: Empirical resistance procedures with limited applicable area
Method Applicable area Author and publication

Ayre Cargo ships Remmers and Kempf (1949)
Taggart Tugboats Taggart (1954)
Taylor-Gertler Slender cargo ships and warships Gertler (1954)
Series-60 Cargo ships Todd (1957)
BSRA Cargo ships Moor et al. (1961)
Helm Small ships Helm (1964)
Danckwardt Cargo ships and trawlers Danckwardt (1969)
Oortmessen Small ships Van Oortmerssen (1971)
Lap-Keller Cargo and passenger ships Lap (1954) and Auf’m Keller (1973)
NPL Small ships Bailey (1976)
Digernes formula Fishing vessels Digernes (1982)
HSVA Catamarans Fritsch and Bertram (2002), Bertram (2012)
According to Bertram (2012), all the systematic series and most of the regression-based methods in Table 3.1
are out of date, and several inaccurately predict the ship resistance. The reason may be the evolution of the
hull form. Therefore, these are disregarded as suitable methods in the current work. However, some more
’modern’ empirical methods with general applicability are widely used today, and these are presented in the
following.
3.1.1 Guldhammer-Harvald
The procedure by Guldhammer and Harvald (1974) in its latest form, including update of procedure by Ander-
sen and Guldhammer (1986) and by Kristensen et al. (2017).
15
Chapter 3. Empirical Prediction Methods
Published: 1965, 1974, 1986 (update of procedure), 2012 (update of procedure)
Area of application: Universal, tankers, single and twin screw vessels
Basis for procedure: Extensive analysis of well-known published model tests such as Lap (1954), Gertler
(1954), Todd (1957), Moor et al. (1961)
Output: Total calm-water resistance RT , thrust factor t, wake fraction w
Key calculations:
CT = CR + CF + CA + CAA (3.1)
p
Where the residual resistance coefficient CR is f (Fn or V / L, L/r1/3 , CP ). CF is found by the ITTC’57 line
and no form factor is applied. The hull-propeller interaction parameters are based on values given in diagrams
in Harvald (1992). Kristensen et al. (2017) presents regression formulas for the diagrams that may be used
for the calculations, including a bulb correction and corrections for t and w.
Input: As defined in accordance with Birk (2019)
Table 3.2: Required and optional input parameters for the Guldhammer-Harvald method
Required parameters Symbol

Length between perpendiculars Lpp
Length in waterline Lwl
Length over wetted surface Los
Max. molded beam in waterline B
Molded draft T
Volumetric displacement 5
Block coefficient CB
Prismatic coefficient CP
Area of ship and cargo above waterline AV T
Propeller diameter Dp
Optional parameters
Longitudinal center of buoyancy LCB
Transverse cross section area of bulb ABT
Wetted surface S
Wetted surface of appendages Sapp
Form factors, fore and aft body FF , FA
Permissible range: As defined in accordance with Birk (2019)
Table 3.3: Recommended range for speed and main dimensions in Guldhammer-Harvald
Parameter Symbol Unit Range

Froude number Fn  0.33
Block coefficient CB [0.55, 0.85]
Length-beam ratio L/B
p [5.0, 8.0]
Length-displacement ratio L/ 3 5 [4.0, 6.0]
Remarks:
• The resistance for ships with small L/B is underestimated (Schneekluth and Bertram, 1998).
References:
Guldhammer and Harvald (1974), Andersen and Guldhammer (1986), Schneekluth and Bertram (1998), Birk
(2019).
16
Chapter 3. Empirical Prediction Methods 3.1 Empirical Resistance Procedures
3.1.2 Holtrop-Mennen
Resistance procedure by Holtrop (1984), often referred to as Holtrop’84.
Published: 1977, 1978, 1982, 1984, 1988 (Update of procedure not applied in the current work)
Area of application: Universal, wide range of ship types
Basis for procedure: Regression analysis of database of the Dutch model basin MARIN
Output: Total calm-water resistance RT , w, t and power prediction
Key calculations: ITTC’78 resistance procedure applied including form factor
RT = (1 + k)RF + RW + RA + RAP P + RAA + RB + RDB (3.2)
RB is resistance due to bulbous bow near the water surface. Coefficients for computing the form factor (1+k)
and the wave resistance RW is added to Appendix C. A viscous resistance coefficient CV is introduced for
the hull-propeller interaction parameters w, t and ⌘H :
(1 + k)RF + RAP P + RA
CV = 1 2
P (3.3)
2 ⇢V (S + i SAP Pi )
Table 3.4: Required and optional input parameters for Holtrop-Mennen’s method

Max. molded beam in waterline B
Molded mean draft, typically T = 12 (TA + TF ) T
Molded draft at aft and forward perpendicular TA , TF
Volumetric displacement r
Prismatic coefficient CP
Midship section coefficient (or CM = CB /CP ) CM
Waterplane area coefficient CW P
Longitudinal center of buoyancy LCB
Area of ship and cargo above waterline AV
Area of immersed transom AT
Area of bulbous bow ABT
Height of center of ABT hB
Stern shape factor Cstern
Optional parameters
Wetted surface S
Half angle of waterline entrances iE
Diameter of bow thruster tunnel dT H
Permissible range: As defined in accordance with Birk (2019)
17
Table 3.5: Recommended range for speed and main dimensions in Holtrop-Mennen

Froude number Fn  0.45
Prismatic coefficient CP [0.55, 0.85]
Length-beam ratio L/B [3.9, 9.5]
Remarks:
• Bulbous bow and transom stern taken into account.

• According to Steen (2020), the bulb correction in Holtrop-Mennen is generally not applied.
• Covers a wide range of ships but require many parameters.
References: Holtrop (1977, 1984), Holtrop and Mennen (1982), Schneekluth and Bertram (1998), Birk
(2019), Bertram (2012), Steen (2020)
3.1.3 Hollenbach
Resistance and power prediction procedure by Hollenbach (1998). The estimation of residuary resistance is
emphasized in the following.
Published: 1997, 1998
Area of application: Universal, modern cargo ships, single and twin screw ships
Basis for procedure: Analysis of database of Vienna ship model basin
Output: Total calm-water resistance RT , w, t and power prediction
Key calculations:
Mean, minimum and maximum total resistance coefficients:
CT,mean = CR,mean + CF + CA + CAP P + CAA

CT,min = CR,min + CF + CA + CAP P + CAA (3.4)
CT,max = h1 CR,mean
The residuary resistance is based on (BT ) instead of wetted surface S . The method does not include a form
factor k . According to Steen et al. (2016) the calculation can be improved by introducing a form factor as
follows:
CT = (CF + CF ) · (1 + k) + CR + CA + CAP P + CAA (3.5)

where the residual coefficient is a function of ship model frictional coefficient CF m as f (Rn , Fn ):
B·T
CR = CR,Hollenbach · k · CF m (3.6)
S
Coefficients of CR,Hollenbach is added to Appendix A. The added frictional resistance is also included as dis-
cussed in Section 2.1. The resistance estimate is combined with open water tests and corrected for the
hull-propeller interaction by:
⌘R - relative rotative efficiency

t - thrust deduction fraction
w - wake fraction
18
Chapter 3. Empirical Prediction Methods 3.2 Added Resistance due to Wind and Waves
Table 3.6: Required and optional input parameters for Hollenbach’s method

Length between perpendiculars Lpp
Length over wetted surface Los
Molded beam B
Molded draft at aft and forward perpendicular T A , TF
Volumetric displacement 5
Area of ship and cargo above waterline AV
No. of appendages (rudders, shaft brackets, bossings, thrusters) Nappendages
Optional parameters
Wetted surface S
Diameter of bow thruster tunnel dT H
Permissible range: As defined in accordance with Birk (2019) for a single screw vessel on design draught
Table 3.7: Recommended range for main dimensions in Hollenbach

(*) Extended range for mean resistance calculation by Birk (2019)

Block coefficient CB [0.49, 0.83]*
Length-beam ratio Lpp /B [4.71, 7.11]
Beam-draught ratio B/T [1.99, 4.00]
Propeller diameter-draught ratio Dp /T
pA [0.43, 0.84]
Length-displacement ratio Lpp / 3 5 [4.49, 6.01]
Remarks:
• Resistance estimated for trial conditions and for a ship without propulsor.
• Based on a relatively modern database.
References:
Hollenbach (1998), Schneekluth and Bertram (1998), Bertram (2012), Birk (2019)
3.2 Added Resistance due to Wind and Waves

Several empirical methods to predict added resistance due to wind and waves are identified in this section.
The methods vary in level of detail and applicable range and will be compared in Section 5.1.
ITTC (2014) recommends four different methods to predict added wave resistance. Two of the methods re-
quire tank tests and are not relevant for fleet calculations. However, the other two are empirical and are also
recommended in ISO 15016 standard by ISO Technical Committee (2015). The respective empirical methods
for added wave resistance predictions are developed by MARIN (2006). These are presented in the following
based on the recommended procedures in the ISO 15016 standard. If limited heave and pitch motions can
be assumed, STAWAVE-1 is applicable, otherwise, STAWAVE-2 is recommended.
In cases where wind tunnel tests are unavailable, ITTC (2014) recommends two empirical methods for pre-
dicting added wind resistance. The methods are developed by MARIN (2006) and Fujiwara et al. (2017) and
presented in the following. In addition, the method by Blendermann (1995) is included. The study is based on
extensive wind tunnel tests and is widely applied in the literature.
19
3.2.1 STAWAVE-1
A simplified method to estimate the added resistance in waves for modern ships when limited input data is
available. Developed by MARIN (2006) and recommended in ISO 15016.
Published: 2006 (MARIN), 2015 (ISO 15016)
Area of application: Universal, present-day ships, limited to low-to-mild sea states (defined under ’Remarks’)
Basis for procedure: Model tests in MARIN’s Depressurised Wave Basin (DWB)
Output: Mean wave resistance in long crested irregular waves, approximated by the mean wave reflection
resistance
Key calculations: Equation 3.7 estimates the mean wave resistance RAW L , in long crested irregular waves:
r
1 2 B
Rwave ⇡ RAWL = ⇢S gH1/3 B (3.7)
16 LBW L
Input: As defined in Table 3.8
Table 3.8: Required input parameters for STAWAVE-1

Significant wave height H1/3
Ship breadth B
Length between perpendiculars LP P
Distance of the bow to 95 %
maximum breadth on the waterline LBW L
Where LBW L is defined according to Figure 3.1 from ISO 15016.
Figure 3.1: Definition of LBW L in STAWAVE-1 (ISO Technical Committee, 2015)
Remarks: Method limited to sea states where the following can be assumed:
p
• Low to mild sea states with restricted wave heights (H1/3  2.25 LPP /100)
• Waves from ahead [0 to ± 45 ( )]
) Limited heave and pitch motions
) Wave reflection dominates the added wave resistance
) Wave induced ship motions can be neglected
References: MARIN (2006), ISO Technical Committee (2015)
3.2.2 STAWAVE-2
An empirical correction method developed by MARIN (2006). The method applies a frequency response func-
tion for ships with heave and pitch, and covers both wave reflection and induced motion effects.
20
Area of application: Universal, present-day ships, within the defined range defined in Table 3.10
Basis for procedure: Model tests in MARIN’s Depressurised Wave Basin (DWB)
Output: Mean wave resistance in long crested irregular waves
Key calculations: Equation 3.8 estimates the mean wave resistance in long crested irregular waves:
Z 1
RAWL (!; VS )
Rwave = 2 2 S⌘ (!)d! (3.8)
0 ⇣A
The mean resistance increase in regular waves RAW L is
RAWL = RAWML + RAWRL (3.9)

where the motion induced resistance component is
2 B2
RAWML = 4⇢S g⇣A raw (!) (3.10)
LPP
and the wave reflection component is
1 2
RAWRL = ⇢S g⇣A B↵1 (!) (3.11)
2
raw (!) is the empirical transfer function as f (Fn , Lpp , kyy , g, CB ). The entire calculation is lengthy and there-
fore not outlined further here.
Table 3.9: Required input parameters for STAWAVE-2

Ship speed V
Ship breadth B
Length between perpendiculars LP P
Draught at midships T
Radius of gyration in lateral direction kyy
Significant wave height H1/3
Wave amplitude ⇣A
Wave number k
Modified Bessel function of first kind of order 1 I1
Modified Bessel function of second kind of order 1 K1
Frequency spectrum (ISSC for wind waves) S⌘
Permissible Range: As defined in Table 3.10
Table 3.10: Permissible range for STAWAVE-2

Wave heading direction deg [ ] [0, ±45]
Length Lpp m [75, 350]
Block coefficient CB [0.50, 0.90]
Length-beam ratio Lpp /B [4.0, 9.0]
Beam-draught ratio B/T [2.2, 5.5]
Froude number Fn [0.10, 0.30]
Remarks:
21
• Wave spectrum defined without forward speed.

• Pierson-Moskowitz spectrum recommended for fully developed sea.
• JONSWAP spectrum recommended for young developing sea states.
3.2.3 STA-JIP wind

An empirical method to predict wind resistance by MARIN (2006), and recommended in ISO 15016. The
method presents wind resistance coefficients for a range of common present-day ship types.
Area of application: Common present-day ship types, including tankers, LNG carriers, container ships, car
carriers, ferry/cruise ships and general cargo ships
Basis for procedure: Systematic wind tunnel experiments and CFD simulations conducted by MARIN (2006),
Witherby & Co (1985) for OCIMF, and Blendermann (1995)
Output: Wind resistance Rwind
Key calculations: Equation 3.12 estimates the wind resistance, based on tabulated values for the wind re-
sistance coefficient CX ( WR )
2
Rwind = 0.5⇢A · CX ( WR ) · AVT · VWR 0.5⇢A · CX (0) · AVT · VG2 (3.12)
Table 3.11: Required input parameters for STAJIP wind

Air density ⇢A
Wind resistance coefficient CX
Relative wind direction at the reference height WR
Transverse projected area above waterline AV T
Relative wind velocity at reference height VW R
Ship speed over ground VG
Remarks:
• Wind resistance coefficients are applicable to tankers, LNG carriers, container ships, car carriers, fer-
ry/cruise ships and general cargo ships.
• Procedure is not applicable to tugs, offshore supply vessels, fishing vessels and fast craft.
3.2.4 Fujiwara
A method to predict wind resistance coefficients by Fujiwara (2005), and recommended in ISO 15016. The
method is extended by Fujiwara et al. (2017) to include regression formulas to estimate geometric parameters
of the ship superstructure.
Published: 2005, 2017 (extension)
22
Basis for procedure: Wind tunnel test data and regression formulas
Output: Wind resistance coefficient CX ( WR )
Key calculations: CX as function of the relative wind direction, regression coefficients, and a number of
geometrical properties listed in Table 3.12, and illustrated in Figure 3.2 from Fujiwara et al. (2017)
Figure 3.2: Definition of structural parameters in Fujiwara et al. (2017)
Table 3.12: Required input parameters for Fujiwara wind resistance coefficients

Relative wind direction at the reference height WR
Length overall Loa
Breadth B
Transverse projected area above waterline AV T
Lateral projected area above waterline AV L
Lateral projected area above upper deck AOD
Distance from midship section to centre of AV L CM C
Height of top of superstructure HBR
Height from waterline to centre of AV L HC
Smoothing range (normally 10 degrees) µ
Remarks:
• The method require a high level of detail for superstructure parameters.
References: Fujiwara (2005), Fujiwara et al. (2017), ISO Technical Committee (2015)
3.2.5 Blendermann
Blendermann (1986, 1994, 1995, 2004) presents an extensive study of wind loads on ships. The method is
extended in later studies to include various wind load coefficients. Here, the method is outlined with regards
to the longitudinal resistance.
Published: 1986, 1994, 1995, 2004
Basis for procedure: Wind tunnel test data and regression formulas
Output: Wind resistance coefficient CX ( WR ), for both lateral and longitudinal resistance, cross-force and
rolling-moment
Key calculations: Tabulated values for the wind resistance coefficient CX ( WR ) as function of the relative
wind direction at the reference height WR . Table 3.13 presents the longitudinal wind resistance coefficients
for head wind (Blendermann, 1994). The complete set of coefficients are tabulated and added to Appendix E.
23
Table 3.13
Type of Vessel CX ( WR = 0)
Car carrier 0.55
Cargo vessel, loaded/container on deck 0.65/0.55
Container ship, loaded 0.55
Drilling vessel 0.60
Ferry 0.45
Fishing vessel 0.70
LNG tanker 0.60
Offshore supply vessel 0.55
Passenger liner 0.40
Research vessel 0.55
Speed boat 0.55
Tanker loaded/ballast 0.90/0.75
Tender 0.55
Input:
Ship type and relative wind direction
References:
Blendermann (1986, 1994, 1995, 2004)
3.2.6 Townsin & Kwon

Townsin and Kwon (1983) presents a study on estimating the influence of weather on ship performance. An
extension to the method by Kwon (2008) presents a formula to estimate the speed loss due to added resis-
tance in wind and waves. Here, the method is outlined in the newest version with the extension. Lu et al.
(2015) presents a modification of Kwon, but the new coefficients are not published and therefore not further
studied here.
Published: 1983, 1993, 2008 (Extension by Kwon)
Area of application: All weather directions are applicable, as head sea and head wind is modified by direc-
tion factors in the procedure
Basis for procedure: Analysis of a wide range of ships having Series 60 forms
Output: The reduced speed in weather due to increased resistance in wind and waves expressed as added
resistance
Key calculations: ✓ ◆
4V
· 100% = C CU CF orm (3.13)
V
Where:
C is the direction angle coefficient as f (BN ), measured with respect to the ship bow
CU is the speed reduction coefficient as f (CB , Fn )
Cf orm is the ship form coefficient as f (shiptype, BN, r)
Let x be the reduction factor:

✓ ◆
4V C CU CF orm
x= = (3.14)
V 100%
The ship brake power PB is computed as follows, where ⌘tot is a function of (V + 4V ).
24
Chapter 3. Empirical Prediction Methods 3.3 Propulsion Prediction Methods
RT · (V + 4V )
PB =
⌘tot
RT · (V + x · V )
=
⌘tot
(3.15)
V · (RT + x · RT )
=
⌘tot
V · (RT + Radded )
=
⌘tot
RT is the calm water resistance and the final added resistance is computed as Radded = x · RT . The method
originally expresses the reduced speed as VW = V 4V . However, if a ship speed of V is observed for a
ship, the calculated added resistance must include V + 4V .
Table 3.14: Required input parameters for the Kwon method

Ship speed V
Beaufort number BN
Length in waterline LW L
Volumetric displacement r
Mean wave and wind direction relative to ship
Remarks:
• Wind-generated waves are assumed.
• Not possible to differ between increased resistance due to waves and increased resistance due to wind.
• In the current work, BN is computed based on the official Beaufort Wind Scale.
References:
Townsin and Kwon (1983, 1993), Kwon (2008).
3.3 Propulsion Prediction Methods

Empirical procedures to determine the propulsive efficiency range from simple empirical estimates to more
detailed propeller design procedures. Primarily three different approaches to determine the propulsive effi-
ciency are suggested and will be outlined in the following.
The first approach is to determine each propulsive efficiency term, i.e., ⌘D = ⌘0 ⌘H ⌘R . This implies finding
the open water efficiency, which require an estimated propeller design. Relevant propeller series to determine
the required parameters are outlined in Subsection 3.3.1.
The second approach is to apply simple empirical formulas requiring limited input for the propulsive efficiency
⌘D directly. Several methods are found in the literature and presented in Section 3.3.2. Section 3.3.3 presents
the third suggested approach, which is to apply the sea-margin to find the propulsive efficiency.
As the empirical prediction methods mainly are valid for calm water conditions, corrections for the load varia-
tions in waves must be applied. Relevant correction methods are presented in Section 3.3.4.
25
3.3.1 Propulsive Efficiencies

Methods to determine each propulsive efficiency term are identified in the following.
Open water efficiency
A wide range of propeller series are identified and outlined briefly in the following, as presented in Birk (2019).
The relevance of the series is reviewed, and those with limited applicable areas are listed in Table 3.15.
Table 3.15: Propeller series with limited applicable area
Method Applicable area Author and publication

Gawn series High speed and naval vessels Gawn (1953)
KCA (Gawn-Burrill) series High speed and naval vessels Gawn and Burrill (1957)
Newton-Rader series Small high speed vessels Newton and Rader (1961)
DTMB skewed series Not applicable for propeller design Boswell (1971)
KA-series Ducted propellers only Oosterveld (1970)
Tillig new propeller series Conventional ships Tillig (2020)
According to Birk (2019), all the listed propeller series listed in Table 3.15 have limited applicability, and some
are out of date. Therefore, these are not further considered in the current work. In addition to the listed pro-
peller series, some series for controllable pitch propellers exist. These also have limited applicability. However,
the Wageningen B-series (Oosterveld and van Oossanen, 1975) covers a wide range of principal propeller
characteristics and has a wide area of applicability. The series represents the most ’modern’ and extensive
procedure. Tillig (2017) suggests the Wageningen-B series is outdated and presents a new propeller series,
developed to account for present propeller designs. However, Kristensen et al. (2017) presents a simplified
method to obtain quick estimates with the Wageningen-B series based on limited input, and it is therefore
selected for further investigation.
Wageningen B-series
The Wageningen B-Series (Troost, 1951) also referred to as Troost series in the litterature, is the most exten-
sive propeller series developed (Birk, 2019). There is a wide data range in terms of combinations of number of
blades Z , expanded area ratio AE /A0 and pitch-diameter ratio P/Dp . Applying the series require a number
of estimated parameters, which is a challenge when the propeller design is unavailable. However, Kristensen
et al. (2017) present a simplified version of applying the method which is outlined in the following.
Approximated values from Wageningen B-series

Breslin and Andersen (1994) present curves for approximated values for ⌘0 from the Wageningen B-series,
denoted as ⌘0,W ag . The efficiency is defined as a function of the thrust load bearing coefficienct CT h as
presented in Equation 3.16
8 RT
CTh = · 2 (3.16)
⇡ (1 t) · ⇢ · (VA · Dp )
Here, the coefficient is independent of the rate of revolutions. RT is the ship resistance and VA = (1 w) · V
is the inflow velocity to the propeller. Kristensen et al. (2017) present a relation between the ideal propeller
efficiency and the approximated values ⌘0,W ag to expand the range for the approximated Wageningen method.
The efficiency can then be estimated as follows, as long as f (CT h ) is not lower than 0.69:
2
⌘0 = ⌘0,W ag = p · f (CT h ) (3.17)
1+ CT h + 1
where
f (CT h ) = 0.81 0.014 · CT h (3.18)
26
Hull efficiency
In order to determine the hull efficiency, the thrust deduction factor and the wake factor must be found, as
Equation 3.19 presents.
1 t
⌘H = (3.19)
1 w
The wake and thrust factor are significantly influenced by the shape of the ship, specifically the aft hull. There-
fore, prediction methods are usually a function of the block coefficient CB or the prismatic coefficient CP .
Other influencing factors regarding the propeller are often not considered. Several empirical estimates with
few input parameters for t and w exist. It is only considered relevant to elaborate on those with general appli-
cability. As identified in Section 3.1, all the calm water resistance methods include predictions of thrust and
wake factors. These are considered to have general applicability and are briefly outlined below.
Guldhammer-Harvald:
Harvald (1992) presents estimates for t and w as outlined by Kristensen et al. (2017), according to Equation
3.20
⇣ ⌘
B Dp
w = w1 L , CB + w2 (form, CB ) + w3 L
⇣ ⌘ (3.20)
B D
t = t1 L , CB + t2 (form) + t3 Lp
where w2 and t2 is equal to zero for the common N-shaped hull form, and DP is the propeller diameter.
The parameters of Equation 3.20 can be approximated in accordance with diagrams from Harvald (1992).
Kristensen et al. (2017) present approximated values from these diagrams by regression formulas. The
calculation is lengthy and therefore added to Appendix B.
Holtrop-Mennen:
Equation 3.21 and 3.22 by Holtrop (1977) represent general formulas for single-screw ships.
L B DP2
t = 0.001979 · + 1.0585 · 0.00524 0.1418 · (3.21)
B (1 CP ) L B·T
 s
LW L c11 CV B
w = c9 c20 CV 0.050776 + 0.93405 + 0.27915c20 + c19 c20 (3.22)
TA 1 CP 1 LW L (1 CP 1 )
The coefficients of Equation 3.22 are tabulated values from Birk (2019), added to Appendix C.
Hollenbach:
Hollenbach (1998) estimates t, w, and ⌘H , somewhat differently from Guldhammer-Harvald and Holtrop-
Mennen. Here, the procedure is outlined according to Birk (2019) for a single screw vessel on design draught.
The thrust deduction factor is assumed to be constant, t = 0.190. The wake fraction and hull efficiency is
determined based on the model scale hull efficiency in Equation 3.23.
✓ ◆ 0.58 ✓ ◆0.1727 ✓ ◆ 0.1334
0.3977 RT mean B DP2
⌘HM = 0.948CB (3.23)
RT T BT
RT is the calm water resistance, while RT mean is the mean calm water resistance in Hollenbach. Then, the
model scale wake fraction is computed according to Equation 3.24
1 t
wT M = 1 (3.24)
⌘HM
Finally the full scale wake fraction is determined by Equation 3.25, according to the ITTC’78 Performance
Prediction Method
27

CF S + CA
w = wT S = (t + 0.04) + (wT M (t + 0.04)) (3.25)
CF M
CF S and CF M are the frictional resistance coefficients for full scale, and model scale, respectively. When w
is calculated, ⌘H is finally determined.
Relative rotative efficiency
As outlined in the theory section, the relative rotative efficiency is influenced by several effects, which makes
it difficult to estimate with few input parameters. However, Holtrop (1977) presents Equation 3.26 for single-
screw ships:
⌘R = 0.9922 0.05908 · AE /A0 + 0.07424 · (CP 0.0225 · LCB) (3.26)
where
AE /A0 - Propeller expanded blade area ratio (EAR)

CP - Prismatic hull coefficient
LCB - Longitudinal centre of buoyancy
The relative rotative efficiency is often close to 1.00 and Alte and Baur (1986) recommend using ⌘R = 1.00 for
single-screw ships as a simple estimate.
3.3.2 Empirical Formulas for the Propulsive Efficiency

There are some complete empirical procedures developed to estimate the propulsive efficiency. Several of
these are only applicable for specific ship types and belong to the methods in Table 3.1. The identified
methods are presented in Table 3.16.
Table 3.16: Empirical Propulsive Efficiency Prediction formulas
Estimation Formula Applicable area Reference

⌘D = 0.836 0.000165 · n · r1/6 Cargo ships and trawlers Danckwardt (1969)
p
⌘D = 0.885 0.00012 · n · Lpp Cargo and passenger ships Auf’m Keller (1973)
p
n Lpp
⌘D = 0.84 10000 General applicability Emersons formula (Watson, 1998)
Here, n is the propeller rpm. According to Watson (1998), Emersons formula is derived for low propeller rpm
although extended to modern propeller design. These ranges define the applicable areas and are illustrated
in Figure 3.3 below. QPC refers to the quasi-propulsive constant i.e. the propulsive efficiency ⌘D .
28
Figure 3.3: Applicable ranges for Emersons formula, inclduing the extension by Watson (1998)
3.3.3 Prediction Based on Sea-Margin

The following method represents a simple way to estimate the propulsive efficiency in ideal conditions, which
can be tuned to fit the fleet better. In ship design, it is common practice to account for wind and waves by
using a sea-margin (SM) on the power, which is typically set to 15% (SFI, 2016). Further, ships operate on
a nominal continuous rating (NCR), which is typically 85 % of the max continuous rating (MCR). Sea-web
provides installed power on the ship, which can be combined with these margins to find the required power in
ideal conditions (i.e., assuming that the 15% sea-margin is applied) as presented in Equation 3.27.
Ps ⇡ (100 15%) · 85% ·Pinstalled (3.27)

| {z } |{z}
SM M CR
By calculating the related calm-water resistance for a ship with speed V , the final ideal propulsive efficiency
can be estimated by modifying Equation 2.18 for required calm-water power as presented in the Theory
Section, to Equation 3.28
RT · V
⌘D ⇡ (3.28)
Ps · ⌘ S
The ship speed can be taken as the service speed provided by Sea-web, and losses in the transmission
system ⌘S must be estimated.
3.3.4 Load Variation Correction

The former methods outlined in this section are mainly valid for calm water conditions and must be corrected
for the load variations on the propeller in a seaway. Three correction procedures are presented in the following.
ISO 15016 correction
A procedure to correct for the change of propulsive efficiency due to added resistance is recommended in the
ISO 15016 standard (ISO Technical Committee, 2015). The method is based on a load variation test and as
presented in Equation 3.29, the efficiency is varying linearly with the increased resistance.
⌘tot R
= ⇠P +1 (3.29)
⌘D RT
⌘tot - Efficiency in trial conditions
⌘D - Efficiency in ideal conditions
R - Added resistance in weather
RT - Calm water resistance
⇠P - Slope of the linear curve in the load variation test
29
The expression for the power correction is presented in Equation 3.30:
Ptot = PS + P ⇣ ⌘
RV ⌘tot (3.30)
P = ⌘tot + PD 1 ⌘D
P - Increased power required due to efficiency loss

Ptot - Delivered power in trial conditions
PS - Delivered power in ideal conditions
Here, ⇠P is the unknown parameter that must be estimated. ⇠P = 0 represent zero correction for changed
efficiency, which is a significant simplification of real conditions. In the ISO 15016 standard test, ⇠P remains at
[0.25,0.24,0.25] for the speed increase of [14.0,15.0,16.0] knots. It may be reasonable to assume a constant
value of ⇠P = 0.25 at this stage.
Modified Wageningen B-series
The approximated Wageningen method by Kristensen et al. (2017) may be modified to include the added
resistance in a seaway. Equation 3.31 calculates the thrust loading coefficient due to calm water resistance
and the added resistance due to wind and waves.
8 (RT + Radded )
CTh = · (3.31)
⇡ (1 t) · ⇢ · (VA · Dp )2
By including the added resistance, CT h is corrected and a corrected propeller efficiency is determined
⌘0,corrected .
Propulsive efficiency by Lindstad et al. (2011, 2013, 2014)
Lindstad et al. (2011, 2013, 2014) present the propulsive efficiency as ⌘tot according to Equation 3.32. The
ideal efficiency ⌘D is corrected for speed reductions and waves in the seastate.
r ! !
V ⇣ ⌘
⌘tot = min ⌘D j+k· , ⌘D 1 r · H 13 (3.32)
Vd
Here, Vd is the design speed. The relation is determined based on the work by Minsaas (2006) and empirical
investigations of propulsion efficiencies by the authors. ⌘D is typically in the range of 0.6 0.7 (Lindstad et al.,
2011). The first term corrects the propulsion efficiency for voluntarily reduced speeds below the design speed.
j and k are empirical constants related by j + k = 1. These are determined based on empirical analysis, and
determine the rate of reduced efficiency for reduced speed. The second term determines the correction for
waves based on the significant wave height of the sea state. r is the factor determining the rate of reduction
and must be found empirically.
3.4 Complete Power Prediction Models

As mentioned in the introduction, new models have been developed for ship powering prediction estimates
with the common objective of fleet-wide emission calculations. Therefore, the models differ from the previously
evaluated empirical methods commonly used in preliminary ship design. Amongst other things, the new
models combine existing empirical methods with new research. An overview of the relevant models is given
in the following.
3.4.1 ShipCLEAN by Tillig et al.

ShipCLEAN by Tillig et al. (2018) is a power prediction method composed of existing empirical formulas and
new developed procedures. The method is applicable to conventional ships. Tillig (2020) presents four ap-
proaches for predicting the power, based on stages of available information. This section outlines ShipCLEAN
30
Chapter 3. Empirical Prediction Methods 3.4 Complete Power Prediction Models
according to stage 1, requiring only main dimensions as input.
The calm water resistance coefficient is decomposed as Equation 3.33
CT = CF + CR (3.33)
where the residual resistance is evaluated according to Hollenbach and the updated Guldhammer-Harvald
method by Kristensen et al. (2017). The valid ranges are checked for both methods and the resistance is
calculated as the mean of the valid methods. For Hollenbach, the mean resistance estimate for a ship on
ballast-draught is applied.
Added resistance due to waves are determined by the average of STAWAVE-2 and a method by Liu and Pa-
panikolaou (2016). The added wind resistance is predicted by Blendermann (1995).
The propulsive efficiency is computed according to Equation 3.34
⌘D = ⌘H · ⌘0 (3.34)
where a new propeller series is developed to determine the propeller efficiency. The hull efficiency is calcu-
lated according to the author’s own estimate, as function of the block coefficient:
⌘H = 1.05 + 0.2 (CB 0.4)
The average of five empirical methods is applied to determine the wake fraction then the thrust factor is
estimated by:
1 t = ⌘H (1 w)
A study on load variations on the propeller due to waves is presented. The author suggests to correct the
propeller efficiency, and presents negligible variations in thrust and wake factors.
The study presents several estimates for hull and propulsion parameters. Two new hull series for slender
ships and ships with higher blockage coefficients are developed and applied to find the wetted surface area.
3.4.2 A Power Prediction model by Kristensen et al.

The power prediction method suggested by Kristensen et al. (2017) is composed of existing empirical formulas
and regression estimates for the input parameters. The resistance calculations are based on the original ITTC
1957 method, decomposed as follows:
RT
CT = CF + CA + CAA + CR = 1 (3.35)
2⇢ ·S·V2
The frictional resistance is estimated by the ITTC’57 friction line and the wave making resistance is determined
by Harvald 1983 method. Effects of wind and waves are included in the service allowance ranging from 20-
30%. The study presents a new method to make the bulbous bow correction, as well as new estimates for
thrust factor and wake fraction. In the analysis it is assumed that the bulbous bow is a function of the same
parameters as Harvalds CR estimate, i.e. f (L/r1/3 , CP , Fn ). A separate analysis is also conducted for the
value of the wetted surface area. The calculation is based on Mumford’s formula:
✓ ◆
r
S = 1.025 · LP P · (CB · B + 1.7 · T ) = 1.025 · + 1.7 · LP P · T (3.36)
T
In the study, a regression analysis is conducted for 129 modern ships of different types and sizes, to tune
the constants of the equation above to fit the main ship types more accurately. The propulsive efficiency is
determined by Wageningen B-series and estimation of propeller characteristics to compute the required input:
⌘tot = ⌘H · ⌘0 · ⌘R · ⌘S (3.37)
31
A regression analysis is conducted to modify the Harvald formulas for wake fraction and thrust deduction to
modern ships. There is not made any corrections for change in propulsive efficiency due to added resistance.
The final power Ptot is then determined by:
✓ ◆
service allowance in %
PE = R T · V · 1 + (3.38)
100
PE
Ptot = (3.39)
⌘tot
3.4.3 STEAM 3 by Jalkanen et al.

STEAM3 by Johansson et al. (2017) and Jalkanen et al. (2012), is a ship traffic emission assessment model,
extended from previous versions (STEAM and STEAM2). The model is similar to the MariTEAM model al-
though the power calculations are evaluated differently. The calm-water resistance is evaluated by Hollenbach
and the ITTC friction line, and added weather resistance is calculated according to Kwon (2008).
1
Ptot ⇡ · (RF + RR ) VW (3.40)
⌘D
The propulsive efficiency is determined by Emersons formula for quasi propulsive constant ⌘D , as function of
propeller rate of revolutions n, and ship length between perpendiculars Lpp :
p
n Lpp
⌘D = 0.84
10000
The study is based on a database including about 30,000 ships with propeller characteristics. For the ships
with unknown propeller diameter, this is estimated by a regression analysis on fraction of draught distances.
With regards to simplifications, there is not applied a correction for the change in propulsive efficiency due to
weather or any roughness corrections to the frictional resistance.
3.4.4 A Power Prediction model by Lindstad et al.

Lindstad et al. (2014) presents an empirical power prediction method according to Equation 3.41.
Ps + Pw + Pa
Ptot = + Paux (3.41)
⌘tot
⌘tot is the propulsive efficiency, including corrections for voluntarily speed reductions, and modified to include
effects of waves in the seastate:
✓ ✓ r ◆ ⇣ ⌘◆
v
⌘tot = min ⌘D j + k · , ⌘D 1 r · H 13
Vd
Ps is the calm water power requirement, computed as function of the total resistance coefficient CT :
⇢ · CT · S · v 3
Ps =
2
Pw is the added power required due to increased resistance in wind and waves, calculated similar to the
STAWAVE-1 procedure in ISO 15016:
2
1 caw · ⇢ · g · H1/3 /2 · B2
Ps = (v + u)
2 L
The required power due to wind Pa is evaluated as the ITTC added resistance due to wind procedure:
2
CDA · ⇢a · AV T · (v + ua )
Pa = v
2
32
Chapter 3. Empirical Prediction Methods 3.5 Methods to Estimate Input Parameters
3.5 Methods to Estimate Input Parameters

Even though empirical methods requiring limited input are emphasized in the current work, some input pa-
rameters must be estimated. This section outlines the parameters considered most important to estimate,
and identifies suitable methods for the calculations. Some of the formulas are rough estimates for preliminary
design, and some are new regression estimates, such as the formulas by Kristensen et al. (2017).
3.5.1 Hull Parameters

Table 3.17 presents various hull parameter estimates. The parameters include the waterline length LW L ,
midsection coefficient CM , waterplane coefficient CW P , and wetted surface area S , which are all important
for the resistance calculations. The applicability of the estimates for the new power prediction model are
evaluated further in Section 5.1.4.
Table 3.17: Estimation formulas for hull parameters

LW L = 1.02LP P Tanker/Bulk Kristensen et al. (2017)
LW L = 1.01LP P Container Kristensen et al. (2017)
LW L = 1.01LP P Ro-Ro Kristensen et al. (2017)
CM = 0.93 + 0.08CB General Schneekluth and Bertram (1998)
CM = 0.80 + 0.21CB Large ships Molland (2011)
CM = 0.78 + 0.21CB Small ships Molland (2011)
CW P = 0.67CB + 0.32 General Molland (2011)
CW P = 0.763(CP + 0.34) Tanker/Bulk/Cargo Bertram and Wobig (1999)
CW P = 3.226(CP 0.36) Container Bertram and Wobig (1999)
CW P = (1 + 2CB )/3 General Papanikolaou (2014)

r
S = 0.99 · T + 1.9 · LW L · T Tanker/Bulk Kristensen et al. (2017)
r
S = 0.995 · T + 1.9 · LW L · T Container Kristensen et al. (2017)
r
S = 0.87 · T + 2.7 · LW L · T Ro-Ro Kristensen et al. (2017)
· (1.2 0.34 · CBW )
r
S = 1.025 · T + 1.7 · LP P · T General Mumfords formula
Estimates to determine the superstructure dimensions are presented in Table 3.18. The superstructure pa-
rameters are important for calculating the air and wind resistance. It is generally difficult to obtain accurate
estimates applicable to all ship types. AV T and AV L is the projected area in the transverse and longitudinal
direction, respectively. Bs is the calculation breadth, hs is the calculation height and Ls is the calculation
length for the superstructure geometry.
33
Table 3.18: Estimation formulas for superstructure dimensions

AV T = Bs · (D T + hs ) Tanker/Bulk Kristensen et al. (2017)
AV L = LOA · (D T ) + L s · hs Conventional ships Tillig et al. (2018)
D = 0.087 · LP P General Bertram and Wobig (1999)
hs = 3 · (no. of decks) + 2 [m] Tanker/Bulk Kristensen et al. (2017)
hs = [11 20.6] [m] Feeder Kristensen et al. (2017)
hs = 24.2 [m] Panamax Kristensen et al. (2017)
hs = [24.2 26.8] [m] Post Panamax Kristensen et al. (2017)
hs = 24 + 2 [m] PCTC (Lpp > 100 m) Tillig (2020)
hs = 15 + 2 [m] Conventional ships (Lpp > 100 m) Tillig (2020)
hs = 12 + 2 [m] Conventional ships (Lpp  100 m) Tillig (2020)
Bs = 30 [m] Tanker/Bulk (B > 30 m) Tillig (2020)
Ls = LP P /2 [m] Ro-Ro Tillig (2020)
Ls = LP P [m] Cruise/Container/PCTC Tillig (2020)
Ls = LP P /7 [m] Conventional ships (LP P  30 m) Tillig (2020)
3.5.2 Propulsion Parameters

Table 3.19 presents estimates for the propeller diameter Dp . It should be noted that the estimates by Kris-
tensen et al. (2017) are based on regression analysis of modern ships, and the estimates by MAN (2018) are
rough rule-of-thumb estimates.
Table 3.19: Estimation formulas for propeller dimensions

Dp = 0.396 · Tmax + 1.30 Tanker/Bulk Kristensen et al. (2017)
Dp = 0.623 · Tmax 0.16 Container Kristensen et al. (2017)
Dp = 0.713 · Tmax 0.08 Ro-Ro Kristensen et al. (2017)
Dp = 0.65 · Tmax Tanker/Bulk MAN (2018)
Dp = 0.75 · Tmax Container MAN (2018)
Dp = 0.85 · Tmax Volume ships, high speed MAN (2018)
34
Chapter 4
The MariTEAM Model
This chapter presents the MariTEAM model, developed by IndEcol at NTNU, an interdisciplinary research pro-
gram specialising in environmental sustainability analysis (IndEcol, 2019). The MariTEAM model calculates
fuel consumption and emissions across ship types, various fuels, engines, weather states, and trade routes.
A bottom-up approach is applied to generate results from individual ships to across the fleet. As illustrated
in Figure 4.1, the model can combine ship technology modules, including fuel type, installed engine power,
speed profile, and route, to generate fleet-level assessments, i.e, fleet-wide emission calculations. The ob-
jective of the chapter is to assess the current model’s data streams and power prediction calculations. Some
parts of the assessment are based on a comprehensive study from the project thesis (Dale, 2020).
Figure 4.1: Modules in the MariTEAM model - From ship to fleet by Bouman et al. (2016)
4.1 Data Input

This section outlines the relevant databases for the fleet-wide power predictions in the MariTEAM model. The
objective is to identify suitable databases and assess the typical data quality. It is expected that inaccurate
data represent the main challenge for data quality, i.e., data containing misspellings, wrong numbers, missing
information, or blank fields.
The input data sources to the MariTEAM model are composed of ship AIS data, technical ship data, and
weather data combined with restricted emission area (ECA) data. As illustrated in the model flowchart in
Figure 4.2, the current model input is ship technicals from Sea-web, speed and location of the vessel from
AIS, and data describing wind and waves from the European Centre for Medium-Range Weather Forecasts
(ECMWF). These data streams are parsed and stored before high-performance computing (HPC) is applied
to complete the ship tracks and assign wind and wave conditions to each ship along its track. The flowchart
illustrate the data streams relevant for the power prediction calculation which is further outlined in Section 4.2.
35
Chapter 4. The MariTEAM Model
Figure 4.2: Flowchart of the MariTEAM power prediction model
As most of the input data analysed in the current work are from Sea-web, this database is outlined in detail.
However, other relevant databases are also briefly assessed.
4.1.1 Sea-web
The IHS Maritime & Trade’ Fairplay database Sea-web™, is the largest commercial maritime ship database,
containing technical and operational data on over 200,000 ships (IHS, 2020). According to IMO, Sea-web is
the official database, and is considered to represent the entire world fleet. Figure 4.3 presents the distribution
of ship types in the fleet in terms of the number of ships (blue), and dwt (grey). As can be seen, general
cargo ships represent the largest segment in terms of the number of ships, and dry bulk carriers represent
the primary carrier of dwt.
36
Chapter 4. The MariTEAM Model 4.1 Data Input
Figure 4.3: Distribution of ship types in the merchant fleet by IHS (2020)
In Sea-web, there are 214 different fields to describe the technical, formal, and operational properties of each
ship. A complete list is added to Appendix D. Among the available fields, a selection of 16 parameters is
considered relevant for the power performance calculations based on the review of empirical methods. These
are presented in Table 4.1 below, and most of these are currently included in the MariTEAM model. The
respective definition of each parameter in Sea-web is also included.
Table 4.1: Relevant Sea-web parameters for the power performance predictions
Parameter Symbol Sea-web definition

Breadth B Moulded breadth, otherwise extreme breadth
Deadweight dwt The load corresponding to maximum summer draught
Depth D From the lowest point on the keel to uppermost continuous deck
Draught T Maximum draught at the summer load load line
Engine Stroke Type - 2 stroke or 4 stroke
Engine rpm N Revolutions per minute
IMO number - Unique number, remains constant during the life of the ship
Lightship weight ldt Light displacement tonnage
Length overall LOA The length overall
Length between perpendiculars LP P The length between perpendiculars
Main engine installed power - Power output of main engines
MMSI number - Identification number for VHF radio communication
Number of decks - Number of non-continuous decks
Service Speed Vd Speed maintained under normal load and weather conditions
Ship Type - Specified by a main category and sub-categories
TEU T EU (20 Foot Equivalent Unit) Capacity
Data quality
The MariTEAM model presents a fleet size analysis based on the 2018 Sea-web data, including about 74,000
ships representing the main fleet segments. First, the data is filtered, outliers are removed, and values are
estimated for missing entries. Outliers are identified as faulty data typically with wrong dimensions or odd
values relative to the trend of its segment. As of Sea-web 2018 data, the MariTEAM model identified 18 ships
37
with deviating characteristics. Individual analyses were conducted for the respective ships by checking the
values against Marine Traffic. This way, some of the deviations were corrected.
Statistics of the coverage of each of the fields of interest are also obtained. The ship displacement (LDT) is
the parameter with the most missing entries among the key physical properties. Of the about 74,000 ships
analysed, 1 % is missing the lightship displacement (ldt) value. Of the remaining fields, auxiliary engine stroke
type has the poorest coverage with 74 % missing entries, i.e., more missing entries than present. Methods
for filling in missing data are presented in Section 4.1.3.
4.1.2 Additional Databases

Additional databases have been reviewed with regards to suitability for the MariTEAM model power predic-
tions. The data from Sea-web may be combined with other ship data sources to achieve improved coverage of
the required calculation parameters. Both Sea/Net by Clarksons (2020) and Lloyds list intelligence by (Lloyd’s,
2020) offer similar resources to Sea-web. However, subscriptions for the databases have not been available
in the current work. For future work, the databases should be further investigated.
EU Fuel consumption reporting
According to the EU-council (2019), a new regulation for reporting fuel consumption is implemented, partly
to align the rules with the IMO global data collection system (DCS) for ship fuel consumption. As of 2019,
shipping companies of the EU member states are obliged to follow the EU MRV (monitoring, reporting, veri-
fication) and the IMO DCS system. This includes monitoring fuel consumption, CO2 emissions, and energy
efficiency from ships undergoing voyages and in port. The first reports are to be submitted in 2019 for EU
MRV and 2020 for IMO DCS. The data from these reportings may be utilised in power prediction calculations
or for validation purposes. A suggestion for future work is to investigate the reports when released.
4.1.3 Methods to Fill In Missing Data

Methods to fill in missing entries from the ship technical database are outlined in the following. First, an
overview of the current filling process in the MariTEAM model is presented. Second, a review of additional
methods are described.
Current filling procedures
In the MariTEAM model, regression rules are established and programmed for each input parameter within
each ship segment. E.g., if the ship segment is ’container’ and the response variable that represents the
missing entry is ’ldt’, then the regression rule selects ’GT’ as the first predictor variable. A second predictor
variable is also defined, which gives the second-best fit for the ’ldt’ response data. This example can be
illustrated by the study on container ships in the MariTEAM model by Ringvold (2017). In the master thesis
study, Ringvold compares potential predictor parameters to estimate ’ldt’. Figure 4.4 presents the response
variable ’ldt’ plotted against six different potential predictor variables.
38
Chapter 4. The MariTEAM Model 4.1 Data Input
Figure 4.4: Scatter plot of ldt against potential predictor parameters by Ringvold (2017)
All the parameters follow this regression pattern except ’Main engine stroke’ and ’Main engine rpm’. ’Main
engine rpm’ is assigned a regression rule based on whether it is 2-stroke or 4-stroke. ’Main engine rpm’ must
be either 2- or 4-stroke and is therefore assigned a value based on whether the predictor is above or below a
specific delimiter value.
4.1.4 AIS Data

AIS is a vessel tracking system, introduced by the IMO International Convention of Safety of Life at Sea
(SOLAS) in 2000 IMO (2020). The system is required to identify ships, improve the safety and efficiency of
navigation, and maritime environmental protection. AIS data is both global and historical. In the present work,
the AIS data is supplied by Kystverket. The following data is generally transmitted from a ship by the AIS
system:
• IMO and MMSI number

• Name
• Type
• Position
• Heading
• Speed
• Size (dimensions)
• Draught
• Weight
The highlighted data are input to the power prediction procedure in the MariTEAM model. The IMO and MMSI
number identifies the ship and connect the Sea-web data. Generally, the ship AIS system transmits MMSI,
position, speed, and heading angle every 2 to 10 seconds depending on the ship speed, and every 3 minutes
if the ship is anchored. The remaining data is usually transmitted every 6 minutes. AIS data is transmitted
by VHF radio signals, which typically have an average range of about 20-30 miles (30-50 km) (MarineTraffic,
2018). When the data is received, it may be partially incomplete or incorrectly formatted. Therefore, the
process of parsing and cleaning the data is vital for data quality.
39
Figure 4.5: The MariTEAM AIS data track completer (Containerships, 2017). Figure courtesy of Radek Lonka, IndEcol,
NTNU.
In the MariTEAM model, the AIS data is initially parsed and filtered such that outliers are removed. Then the
post-processing of filling in missing data points is run on high-performance computers. This ship track com-
pleter fills any gaps by interpolating between AIS location data. Figure 4.5 illustrate the completed AIS data
for container ships in 2017. There is a trade-off between computational time and accuracy of the predicted
route for the track completer. Port calls provide data about the ship arrival and departure from ports, which is
applied to help generating the most likely ship route.
ITU List V
ITU is the UN specialized agency for information and communication technologies (ITU, 2018). The ITU list V
is a list of ship stations and maritime mobile service identity assignments (MMSI numbers). It contains all ships
designated with an MMSI number, which is an internationally standardized number for the vessel. It identifies
the ship and is programmed into the AIS and VHF systems onboard. Therefore, the MMSI is considered the
key parameter connecting ship data from Sea-web and AIS. Since the MMSI number is missing for some
Sea-web entries, ITU list V can be used to fill in the missing numbers. According to the MariTEAM model,
about 20% of the MMSI numbers are missing in Sea-web and filled in from ITU list V.
4.1.5 Weather Data

The weather data in the MariTEAM model include historical wave and wind data acquired by the ECMWF
(2014) reanalysis dataset ERA-Interim. ERA-Interim supplies hindcast weather data, continuosly updated in
real time. The data variables for waves and wind are included in Table 4.2
Table 4.2: ECMWF Weather Data Variables
Variable Temporal resolution

Significant wave height of combined wind waves and swell [m] 6 hours
Mean wave period [s] 6 hours
Mean wave direction 6 hours
Northward wind speed (10 m above sea surface) [m/s] 3 hours
Eastward wind speed (10 m above sea surface) [m/s] 3 hours
The weather data is combined with the AIS data of the ship to obtain the relative direction angle and velocity
between ship and weather components.
4.2 Power Prediction Procedure

The MariTEAM model applies empirical methods to develop ship power profiles. Since the methods require
parameters which are not usually available from commercial ship technical databases, some simplifications
and assumptions are made and elaborated on in the following.
40
Chapter 4. The MariTEAM Model 4.2 Power Prediction Procedure

The total calm water resistance is estimated by Equation 4.1
RT = RF + RR (4.1)
where RF is the frictional resistance and RR is the residual resistance as defined in Section 2.1. The frictional
resistance is determined according to the ITTC’57 friction line:
⇢
RF = C F V 2 S (4.2)
2
where
0.075
CF = 2 (4.3)
(log(Rn ) 2)
And the wetted area S is defined by the Mumford formula:
5
S = 1.7 · L · T + (4.4)
T
The residual resistance is evaluted according to Hollenbach (1998), assuming the ’minimum resistance’ hull
form. The related residual resistance coefficient is then:
✓ ◆a1 ✓ ◆a2 ✓ ◆a3 ✓ ◆a4
T B Los Llwl
CR = CR,Standard · · · · (4.5)
B L Llwl L
where CR,Standard is determined by the Froude number and a number of tabulated coefficients which can be
found in Appendix A. The total resistance is evaluated without including a form factor and without accounting
for roughness effects. It is further assumed that the resistance from air and from moving in shallow water are
small relative to the total resistance and is neglected in the resistance calculations.
4.2.2 Added Resistance in Wind and Waves

The added resistance effects from weather is accounted for by adjusting the speed in accordance with the
method(s) presented by Townsin and Kwon (1983) and the Kwon (2008) extension. The weather data input is
from ERA-Interim, including significant wave height of combined wind waves and swell, and magnitude of 10
m wind speed.
✓ ◆
4V
· 100% = C CU CF orm ,
V
(4.6)
VW = V 4V
4.2.3 Propulsive Efficiency

K is the propulsive efficiency defined by Lindstad et al. (2011).
1 1
K= · ⇣ p ⌘ (4.7)
⌘tot ⌘ j + k V /Vd
The propulsion efficiency factor K is a function of the vessel speed V and ⌘ gives the efficiency at the design
speed Vd . ⌘ is typically in the range of 0.6 0.7. The MariTEAM model applies the following constant values
⌘ = 0.65, j = 0.8, k = 0.2.
41
4.2.4 Total Power

The total brake power is finally determined by:
1
PB = · (RF + RR ) VW (4.8)
⌘tot
where VW is the speed over ground corrected for added resistance due to wind and waves according to
Townsin and Kwon (1983); Kwon (2008). If the vessel is a containership that exclusively carries refrigerated
containers (a reefer), additional power required is calculated and added to the total without accounting for
propulsion efficiency.
4.2.5 Assumptions
This section elaborates on the simplifications and unique assumptions made in the power calculations.
Resistance contributions
The calm water resistance is calculated as the sum of frictional and residual contributions. Other contributions
such as form factor, roughness, air resistance, shallow water, base drag (etc.) are considered small relative
to these and therefore neglected.
Hollenbach
The residual resistance is evaluated assuming the ’minimum resistance’ hull form as defined by Hollenbach.
The range of validity for this assumption in terms of ship characteristics are presented below in Figure 4.6. In
Section 5.1 the range is checked against ship data to investigate the validity.
Figure 4.6: Range of validity for minimum resistance in Hollenbach’s method by Schneekluth and Bertram (1998)
Propulsive efficiency
The propulsive efficiency is evaluated as suggested by Lindstad et al. (2011). In addition to the calm water
propulsive efficiency, the formula includes changes in propulsive efficiency due to voluntarily speed reduction.
However, effects of added resistance due to wind and waves are not accounted for in the calculations. This
can be included by the extended formula by Lindstad et al. (2014) or the ITTC method recommended in the
ISO 15016 standard.
42
Chapter 5
New Performance Prediction Method
This chapter presents a new performance prediction method, composed of the most suitable empirical proce-
dures identified in Chapter 3. The objective of the new method is to improve the power prediction procedure
for the current MariTEAM model. In order to achieve a suitable method for fleetwide calculations, calculation
procedures with general applicability is emphasized. The selected calculation procedures applied to the new
model will be described in detail in the following.
5.1 Selection of Methods

In Chapter 3, a wide range of methods are identified, and the following section narrows the selection down
to the most applicable. Several criteria are considered important in the selection of calculation methods for
resistance and propulsion factors. An ideal method applies to a wide range of ship types, provides highly
accurate results, and requires limited input. Further, it should be computationally simple to program. These
requirements serve as the basis for selecting the calculations methods for the new program.

The literature review of empirical resistance methods presented in Chapter 3 identifies three methods with
general applicability; Guldhammer-Harvald, Holtrop-Mennen, and Hollenbach. All methods present regression-
based formulas based on extensive analysis of model tests and are widely applied today.
The applicability of each method is defined in terms of dimensionless ship characteristics. Guldhammer-
Harvald and Hollenbach include more limitations than Holtrop-Mennen. However, all three are expected
to apply for conventional merchant ships. An assessment of the fleet-wide applicability is conducted for the
current MariTEAM database, including 73,000 ships. Each ship in the fleet is checked against the requirement
of each method, and Figure 5.1.1 presents the percentage of ships in the fleet valid to analyse.
Table 5.1: Number of ships in the fleet [%] passing the re-
quirement of each method, all three methods, one of the
three methods, or either Hollenbach or Holtrop-Mennen.
Method Applicability [%]

Hollenbach 42.2 %
Holtrop-Mennen 72.2 %
Guldhammer-Harvald 50.2%
HB & HM & GH 39.2 %
HB | HM | GH 72.8 %
HB | HM 72.8 %
HB | GH 53.4 %
43
Chapter 5. New Performance Prediction Method
As seen in Figure 5.1.1, Holtrop-Mennen covers the broadest range of the fleet by 72.2 %. The combination
of applying either Holtrop-Mennen or Hollenbach gives the most extensive coverage of 72.8 %. Only half
of the fleet can be analysed if either Hollenbach or Guldhammer-Harvald is applied alone. Therefore, it is
considered relevant to include more than one of the methods in the new model.
Hollenbach’s Ph.D. thesis presents a comparison of the accuracy of the three methods when applied to a
wide range of ships (Hollenbach, 1997). Figure 5.1 from Steen et al. (2016) presents the results in terms of
mean deviation and standard deviation of each method. All methods provide the most accurate results for
single-screw vessels on design draught, and the variability increases on ballast draught. Overall, Hollenbach
gives lower standard deviation relative to the others.
Figure 5.1: Comparison of mean and standard deviation of the total resistance predicted by Holtrop-Mennen,
Guldhammer-Harvald and Hollenbach by Hollenbach (1997), retrieved from Steen et al. (2016)
The accuracy of the methods is expected to be dependent on the ship type analysed, since the methods are
regression-based on model tests. As the oldest method, Guldhammer-Harvald is based on the oldest hull
forms (model tests conducted before 1961) and may be expected to provide the least accurate results. How-
ever, the method has been revisited by Kristensen et al. (2017) and is updated for modern tankers, bulk car-
riers, and container ships. The update includes a bulb correction, and according to Schneekluth and Bertram
(1998), modern bulbs may decrease the resistance up to 15-20 %. With this correction, Guldhammer-Harvald
is considered relevant to include in the new method.
In terms of required input, Guldhammer-Harvald represents the simplest method. Both Holtrop-Mennen and
Hollenbach require a considerable number of input parameters unavailable from Sea-web, which must be esti-
mated by other methods. The accuracy of the methods is expected to decrease with the number of estimated
input parameters. However, both Tillig (2020) and Kristensen et al. (2017) present several modern parameter
estimation formulas that may be applied. Regarding computational simplicity, there are negligible differences
between the three methods. Guldhammer-Harvald is based on diagrams of length-displacement ratios, which
Kristensen et al. (2017) presents regression formulas for.
The three methods’ assessment can be summarised in terms of the main criteria; applicability, accuracy, and
required input. Holtrop-Mennen provides the broadest applicable range, which makes it relevant to include.
Based on Hollenbach’s study, Holtrop-Mennen and Hollenbach appear to be more accurate than the original
Guldhammer-Harvald. According to Birk (2019), Hollenbachs’ study indicate that the method is the most re-
liable of the three. Both Hollenbach and Holtrop-Mennen are therefore suitable for the new model. However,
since Guldhammer-Harvald is updated by Kristensen et al. (2017) to fit modern ships, it is considered relevant
to include as well. Without any other clear indication of which method is best, all three are included for further
analyses and a mean estimate of the three methods is computed. Whether the final power prediction benefit
from applying the mean of all three methods will be assessed in Chapter 6 Results and Validation, and further
discussed in Chapter 7.
Total resistance decomposition
In Chapter 2, the resistance components are reviewed according to MARINTEK’s procedure. The most im-
portant resistance contributions are the residuary resistance and the viscous (frictional) resistance. Further,
the roughness correction is expected to be significant, and air resistance may be significant for fast ships.
In addition, the correlation coefficient depends on the selected method; Hollenbach, Holtrop-Mennen, or
44
Chapter 5. New Performance Prediction Method 5.1 Selection of Methods
Guldhammer-Harvald. Therefore, these components are included in the new model, decomposing the total
resistance coefficient as presented in Equation 5.1
CT = CR + (1 + k) (CF + CF ) + CAA + CA (5.1)
Following this decomposition, resistance contributions from base drag and appendages are neglected. These
contributions require detailed input parameters for the transom and appendages which are not available from
Sea-web. The simplification is not expected to be significant for the results. Further, shallow-water effects are
not considered.
The calculation of total calm water resistance follows this decomposition for Hollenbach, Holtrop-Mennen, and
Guldhammer-Harvald. The frictional resistance is computed by the ITTC’57 correlation line:
0.075
CF = 2 (5.2)
(log(Rn ) 2)
The air resistance is calculated according to the ITTC’78 procedure:
⇢air · AV T
CAA = CDA (5.3)
⇢·S
where CAA is defined by Table 5.2 for container ships, tankers and bulk carriers as suggested by Kristensen
et al. (2017). For other ships, the air drag coefficient CDA is determined by Blendermann (1994) according to
Table 3.13 or set to the default value of 0.8 as recommended by Birk (2019). The transverse projected area
above the waterline AV T is estimated by parameter estimates, further outlined in Section 5.1.4.
Table 5.2: Air resistance coefficient values for container ships, tankers and bulk carriers as recommended by Kristensen
et al. (2017)
Ship type CAA · 1000 [-]

Container ships 0.28·T EU 0.126
Small tanker/bulk 0.07

Handysize tanker/bulk 0.07
Handymax tanker/bulk 0.07
Panamax tanker/bulk 0.05
Aframax tanker/bulk 0.05
Suezmax tanker/bulk 0.05
VLCC 0.04
The residual resistance, roughness correction, form factor and correlation allowance are defined somewhat
differently in each method, and the following sections outline the details.
Hollenbach
The total resistance and the hull-propeller interaction parameters in Hollenbach are computed for single-screw
vessels on design draught. This is a simplification, as parts of the fleet comprise twin-screw vessels and ves-
sels on ballast draught. It is assumed to be a reasonable simplification at this stage, but methods to correct
the varying loading conditions are discussed in Section 7.
In the new model, both the minimum and the mean resistance estimate by Hollenbach is included. The mean
resistance prediction RT,mean is considered most relevant and is included in the averaged prediction in the
new model together with Holtrop-Mennen and Guldhammer-Harvald. However, the current MariTEAM model
applies the minimum resistance estimate RT,min , and it is therefore included for comparison and further
investigation in the case study. The total resistance coefficient is calculated as follows (for both the mean and
the minimum estimate):
CT = (CF + CF ) · (1 + k) + CR + CA + CAA (5.4)
45
where CA = 0.06 · 103 (Birk, 2019). The residuary resistance is based on (BT ) instead of wetted surface
S . The method is modified by including a form factor k according to MARINTEK (2020), adjusted for slender
ships according to Steen et al. (2016):
3
k = k0 + k1 = 0.6 · + 75 · (5.5)
where
CB p
= · B · (TA + TF ) (5.6)
LW L
The residual coefficient is further modified as a function of ship model frictional coefficient CF M :
B·T
CR = CR,Hollenbach · k · CF M (5.7)
S
CR,Hollenbach is determined based on tabulated coefficients depending on the mean or minimum resistance
estimate. The coefficients are added to Appendix A. Further, the roughness correction CF is calculated
according to MARINTEK (2020), for H = 150 [µm]:
⇥ ⇤
CF = 110 · (H · V )0.21 403 · CF2 (5.8)
The hull-propeller interaction parameters comprise the thrust factor and the wake fraction. In Hollenbach,
the thrust factor for single screw ships on design druaght is a fixed value t = 0.190. The wake fraction is
determined by the hull efficiency at model scale ⌘HM according to Equation 5.9
✓ ◆ 0.58 ✓ ◆0.1727 ✓ ◆ 0.1334
0.3977 RT mean B DP2
⌘HM = 0.948CB (5.9)
RT T BT
where RT is either RT,mean or RT,min . Further, the model scale wake fraction is determined as Equation
5.10
1 t
wT M = 1 (5.10)
⌘HM
Finally the full scale wake fraction is determined by Equation 5.11

CF + CA
w = (t + 0.04) + (wT M (t + 0.04)) (5.11)
CF M
The hull efficiency is then computed according to Equation 5.12
1 t
⌘H = (5.12)
1 w
When all parameters are computed, the Froude number limitation is applied. In Hollenbach, the valid Froude
number range depend on coefficients and CB .
Holtrop-Mennen
The total resistance and the hull-propeller interaction parameters in Holtrop-Mennen are computed according
to Holtrop’84 for single-screw vessels. Equation 5.13 presents the total resistance:
RT = (1 + k)RF + RW + RA + RAA (5.13)

The calculation of the resistance components are lengthy and include a number of coefficients, these are
therefore added to Appendix C. Only the key calculations are outlined here. The wave resistance RW is a
function of Froude number, subdivided into three sections according to Equation 5.14 (Birk, 2019)
8 ⇥ ⇤
>
> c1 c2 c5 ⇢gV exp m1 F nd + m4 cos F n 2 if F n  0.4
>
<
RW (F n) = Interpolation if 0.4 < F n  0.55 (5.14)
>
>
>
: ⇥ ⇤
c17 c2 c5 ⇢gV exp m3 F nd + m4 cos F n 2 if F n > 0.55
46
where the coefficients are found in Appendix C. c2 depend on the height of centre of bulb area hB , which
is assumed to be 0.6TF based on Birk (2019). The form factor k is computed according to Holtrop’s own
formula, also added to Appendix C. RA includes the effect of hull roughness equal to H = 150 [µm] and is
computed by Equation 5.15
1 2 h X i
RA =⇢V (CA + CA ) S + SAP P (5.15)
2
where SAP P is neglected in the current work. As discussed in Section 3.1.2, the bulb correction is also
neglected. A viscous resistance coefficient CV is introduced for the hull-propeller interaction parameters w,
t, and ⌘H :
(1 + k)RF + RAP P + RA
CV = 1 2
P (5.16)
2 ⇢V (S + i SAP Pi )
The equations for w, t, and ⌘H are previously presented in Section 3.3.1. When all parameters are computed,
the Froude number limitation is applied. In Holtrop-Mennen, the valid Froude number range is [0,0.45].
Guldhammer-Harvald
The total resistance and the hull-propeller interaction parameters are computed according to the updated
Guldhammer-Harvald procedure by Kristensen et al. (2017). Equation 5.17 presents the total resistance
coefficient, where the correlation factor CA includes the effects of roughness of the ship hull.
CT = CR + CF + CA + CAA (5.17)
The residuary resistance CR is determined according to diagrams by Guldhammer-Harvald and corrected for
B/T ratio, hull form and bulb. A correction for LCB not being placed amidships is neglected in the current
work. Equation 5.18 outlines the calculation of CR , where the coefficients of CR, Diagram is added to Appendix
B.
CR = CR, Diagram + CR,B/T6=2.5 + CR,LCB + CR, form + CR, bulb (5.18)
where
✓ ◆
B 3
CR,B/T6=2.5 = 0.16 · 2.5 · 10 (5.19)
T
if B/T 6= 2.5. And the hull form correction is applied for ships with aft or fore body with extremely U or V form:
CR, form :
Fore body Extreme U : 0.1 · 10 3
Extreme V : +0.1 · 103 (5.20)
After body Extreme U : +0.1 · 10 3
Extreme V : 0.1 · 10 3
In the current work, it is assumed that modern tankers/bulk carriers with large CB (> 0.85) have U-shaped
fore body, and fast container ships (Vd > 20knots) have V-shaped entrances. Equation 5.21 and 5.22 present
the bulb correction formula for tankers/bulk carriers and container ships, respectively:
CR, bulb = max( 0.4; 0.1 1.6 · Fn) (5.21)
CR Harvald no bulbous bow

CR, bulb = (250 · F n 90) · (5.22)
100
The hull-propeller interaction parameters t, w, and ⌘H are computed as presented in Section 3.3.1. A common
N-shaped hull form is assumed for the calculations in the current work. When all parameters are computed,
the Froude number limitation is applied. In Guldhammer-Harvald, the valid Froude number range is [0,0.33].
47
5.1.2 Added Resistance in Wind and Waves

Chapter 3 identifies a range of empirical methods to predict the added resistance in wind and waves. Table
5.3 summarises the the identification and highlights the selected methods for the new model.
Table 5.3: Selected methods to predict the increased resistance in wind and waves
Method Added resistance parameter Reference

STAWAVE-1 Wave resistance MARIN (2006)
STAWAVE-2 Wave resistance MARIN (2006)
STA-JIP wind Wind resistance MARIN (2006)
Fujiwara Wind resistance Fujiwara et al. (2017)
Blendermann Wind resistance Blendermann (1994)
Townsin & Kwon Wave and wind resistance Townsin and Kwon (1983), Kwon (2008)
The current MariTEAM model applies Townsin & Kwon for the resistance increase in wind and waves. The
method is widely applicable but does not differ between wave resistance and wind resistance. In order to
obtain a more accurate prediction, methods to predict wave resistance and wind resistance independently are
added to the new model. The selected methods are ellaborated on in the following. Townsin & Kwon is also
included to compare the performance of the various methods.
Wave resistance methods
Both the STAWAVE methods estimate the added resistance in waves for modern ships when limited input
data is available. The methods are recommended in ISO 15016 and are generally applicable to all ship types.
However, STA-1 is limited to mild sea states and returns zero resistance in the new model if the wave height
is above a required limit. STA-2 is therefore included to increase the applicability in higher sea states. In a
study of 10 ships, MARIN (2006) presents a comparison between STA-1, STA-2, and other existing methods
to predict added resistance in waves. The computed results are compared to model test measurements and
indicate that both STAWAVE methods provide a more reliable prediction of added resistance in bow wave trial
conditions, relative to the other methods. Figure 5.2 illustrate the results for a 174 m tanker on various loading
conditions.
Figure 5.2: Comparison of the STAWAVE methods and irregular wave model tests for a 174 m tanker (MARIN, 2006)
A limitation of STA-1 is the required input parameter LBW L , defining the distance of the bow to 95 % of the
maximum breadth on the waterline, as illustrated in Figure 5.3. The length depends on the fore hull shape and
is difficult to estimate based on the limited input available from Sea-web. In the current work, comparison ships
with known LBW L are applied to express the length as a fraction of LP P for the case study. A suggestion for
48
further work is to develop a parameter estimate based on typical hull forms within the ship types, and this is
further discussed in Chapter 9.
Figure 5.3: Definition of LBW L in STAWAVE-1 (ISO Technical Committee, 2015)
STA-2 depends on a defined wave spectrum, and MARIN (2006) recommends applying Pierson-Moskowitz
for fully developed seas or JONSWAP for young developing seas. The most suitable spectrum depends on
the vessel route and will vary within the fleet-wide calculations. For coastal waters, JONSWAP is commonly
applied. In the current work, it is assumed that the majority of ships travel in deep sea with a fully developed
sea state and Pierson-Moskowitz is applied.
Wind resistance methods
ISO 15016 recommends two empirical methods to predict wind resistance; STA-JIP wind and Fujiwara. Both
methods are therefore expected to provide reliable predictions within the applicable ship types. While STA-
JIP wind only requires AV T as input, Fujiwara require a high level of detail for superstructure parameters.
Although Fujiwara et al. (2017) presents regression formulas to estimate the required geometric parameters,
the number of estimated parameters is expected to reduce the accuracy of the results. As presented in Figure
5.4, the two methods apply to five segments each, and it is therefore deemed sufficient only to include one.
As the most straightforward method, STA-JIP is selected. STA-JIP suggests a method to calculate the relative
wind velocity vector, which is neglected in the current work since the parameter is estimated by hindcast
weather data. By including Blendermann (1994), the applicable range of ship types is significantly increased.
Blendermann’s method is widely applied in the literature and is expected to provide reliable predictions for the
wind resistance coefficients. The current work includes coefficients for the ship types added to Appendix E. It
is assumed that the wind resistance for tankers also apply to bulk carriers.
Table 5.4: Comparison of the applicable area for the wind resistance methods
Ship type Blendermann STAJIP-wind Fujiwara

Bulk carrier x
Car carrier x
Cargo vessel x x
Container ship x x x
Drilling vessel x
Ferry x
Fishing vessel x
LNG tanker x x x
Offshore supply vessel x
Passenger liner x x x
Research vessel x
Speed boat x
Tanker loaded/ballast x x x
Tender x
5.1.3 Propulsion Factors

Section 3.3 presents various approaches to determine the propulsion factors and these are narrowed down to
the selected procedure in the following. Propeller information is missing in Sea-web. Therefore, the option of
49
applying a method requiring a significant amount of input parameters is unachievable. Without any other infor-
mation, single screw vessels are considered, following the same assumption as for the resistance calculations.
The current MariTEAM model applies a simple empirical formula by Lindstad et al. (2011) to calculate the
propulsive efficiency. The procedure requires several estimated parameters, such as the propulsive efficiency
in calm water, and empirical constants. In the current work, accurate estimates for these parameters are
not identified, and the method is therefore not implemented. Other existing formulas for ⌘D are presented in
Section 3.3.2 and are deemed too simple for the calculations. Further, a prediction based on sea-margin is
assessed and provides a quick estimate of ⌘D . However, it does not capture load variations on the propeller
due to increased resistance in weather, and therefore not considered suitable for the new model.
A suggested approach is to determine each propulsive efficiency term, i.e., ⌘D = ⌘0 ⌘H ⌘R . This implies finding
the open water efficiency, which requires an estimated propeller design. Kristensen et al. (2017) present an
updated and approximated procedure to apply the Wageningen B-series when only the propeller diameter is
known. A number of parameter estimates for the propeller diameter are identified in Section 3.5.2, enabling
to apply this procedure in the new model. Assuming the calm water resistance methods are valid, the hull
efficiency is determined by these. This procedure is further suitable for the calculations as it accounts for the
varying load on the propeller. Equations 5.23-5.26 present the calculation:
2
⌘0 = p · f (CT h ) (5.23)
1+ CT h + 1
where
f (CT h ) = 0.81 0.014 · CT h (5.24)

and the thrust load bearing coefficient CT h is computed for the total resistance, including added resistance:
8 Rtot
CTh = · 2 (5.25)
⇡ (1 t) · ⇢ · ((1 w)V · Dp )
where w and t are determined by Guldhammer-Harvald, Holtrop-Mennen or Hollenbach. The hull efficiency
is then determined:
1 t
⌘H = (5.26)
1 w
It is further assumed a constant rotative efficiency ⌘R = 1.0 and a constant value of losses in the transmission
system ⌘S = 0.98. Effects of gearbox are therefore neglected.
5.1.4 Input Parameter Estimates

Sea-web provides a limited number of ship characteristics, and empirical formulas and regression estimate
the missing required hull and propulsion input parameters. Based on the selected methods for the new model,
a set of estimated input parameters is selected. The selection comprises modern estimates when available,
ship-type specific estimates, and some assumptions. Propeller diameter estimates are selected based on the
maximum draught, as this is the draught available in Sea-Web. Assumptions include equal trim fore and aft, no
submerged transom and neglected wetted surface of appendages. The significance of these simplifications
is further discussed in Chapter 7. Tables 5.5 - 5.7 present the estimation formulas for hull dimensions and
propeller dimensions.
50
Table 5.5: Estimation formulas for superstructure dimensions
Estimation Formula Applied to Reference

AV T = B · (D T + hs ) General Assumption
hs = 3 · (no. of decks) + 2 [m] General Assumption
hs = [11 20.6] [m] Feeder Kristensen et al. (2017)
hs = 24.2 [m] Panamax Kristensen et al. (2017)
hs = [24.2 26.8] [m] Post Panamax Kristensen et al. (2017)
Table 5.6: Estimation formulas for hull dimensions

LW L = 1.01LP P Container Kristensen et al. (2017)
LW L = 1.01LP P Single screw Ro-Ro Kristensen et al. (2017)
LW L = 1.02LP P Other Assumption
TF = T General Assumption
TA = T General Assumption
(ldt+dwt)103
5= ⇢ General Universal
r
CB = Lpp ·B·T General Universal
r
S = 0.99 · T + 1.9 · LW L · T Tanker/Bulk Kristensen et al. (2017)
r
S = 0.995 · T + 1.9 · LW L · T Container Kristensen et al. (2017)
r
S = 0.87 · T + 2.7 · LW L · T · (1.2 0.34 · CBW ) Single screw Ro-Ro Kristensen et al. (2017)
r
S = 1.025 · T + 1.7 · LP P · T Other (general) Mumfords formula
Sapp = 0 General Assumption
SB = 0 General Assumption
CM = 0.93 + 0.08CB General Schneekluth and Bertram (1998)
CP = CB /CM General Universal
CW P = 0.763(CP + 0.34) Tanker/Bulk/Cargo Bertram and Wobig (1999)
CW P = 3.226(CP 0.36) Container Bertram and Wobig (1999)
CW P = (1 + 2CB )/3 Other (general) Papanikolaou (2014)
51
Table 5.7: Estimation formulas for propeller dimensions

Dp = 0.396 · Tmax + 1.30 Tanker/Bulk Kristensen et al. (2017)
Dp = 0.623 · Tmax 0.16 Container Kristensen et al. (2017)
Dp = 0.713 · Tmax 0.08 Ro-Ro Kristensen et al. (2017)
Dp = 0.5 · ((0.396 · Tmax + 1.30) + (0.623 · Tmax 0.16)) Cargo Assumption
Dp = 0.70 · Tmax Other Assumption
np = 1 (no. of propellers) General Assumption
5.1.5 Potential Improvements to the Current MariTEAM model

Table 5.8 summarises the potential improvements to the current MariTEAM model. First, a filter to identify the
valid methods for a specific ship is implemented in all parts of the calculation. Both the calm water resistance
methods and most of the added resistance methods apply to vessels with dimensions within a limited range.
Further, the calm water resistance is determined by three methods, extending the applicable range for the
current model. Two additional methods are implemented for the added resistance in both wind and waves,
which is expected to provide more accurate results. Finally, the propulsive efficiency is determined by an
extensive propeller series, which is expected to provide a more reliable prediction than the current empirical
formula.
Table 5.8: Potential Improvements to the Current MariTEAM model
Calculation part Current model New model

Filter for identifying valid methods No filter Filter for all methods
Calm water resistance One method Three methods (and mean)
Added resistance in One method combining Two additional methods for wind
wind and waves wind and waves Two additional methods for waves
Propulsive efficiency One empirical method Applying the most extensive

propeller series as of today
5.1.6 Limitations
This section outlines the main limitations in the new model. The significance of these are further discussed in
Chapter 7.
Calm water resistance
• Neglected resistance contributions: Shallow water, wetted transom and appendages
• Single screw ships on design loading is assumed
• Sea-web only provides maximum draught on summer load line
• Even with three methods, 30% of the fleet is not valid to analyse
• The modified Guldhammer-Harvald bulb-correction is only updated for tankers/bulk carriers and con-
tainer ships
52
Chapter 5. New Performance Prediction Method 5.2 Program Structure
Added resistance in wind and waves
• Both STAWAVE methods are only valid for head sea [0 , ±45 ]
• STA-1 input parameter LBW L is estimated by comparison ships
• STA-2 wave spectrum assumes fully developed sea
• Wind resistance accuracy depend on the estimated transverse superstructure area
Propulsive efficiency
• Single screw ships is assumed
• Propulsion factors strongly depend on the estimated propeller diameter
• Rough assumption of constant values for ⌘R = 1.0 and ⌘S = 0.98
• Direct drive assumed and gearbox losses are neglected
5.2 Program Structure

This section is dedicated to explaining how the new model is implemented in MATLAB and the function of
each calculation part. The code is added to Appendix F. Figure 5.4 illustrate the overall structure of the
program, composed of five main modules. First, missing input parameters are estimated. Then the calm
water resistance is calculated, followed by the added resistance in wind and waves. The propulsive efficiency
is calculated as a function of the total resistance before the final brake power requirement is computed. The
details of each of the five modules are elaborated on in the following.
Figure 5.4: Conceptual flowchart of the new program
53
5.2.1 Modules
The details of the five main modules are listed below, and a flowchart, including all functions, are presented in
Section 5.2.2. Note that HB, HM, and GH are abbreviations of Hollenbach, Holtrop-Mennen and Guldhammer-
Harvald, respectively.
Module 1: Estimate missing input
• Estimate missing hull parameters in accordance with Table 5.6

• Estimate propeller diameter based in accordance with Table 5.7
Module 2: Calm water resistance
• Check requirements and determine valid calm water methods (HB, HM or GH)
• Compute the resistance, thrust factor, and wake fraction for all valid methods
• Compute mean values from valid methods: RT , t, w
Module 3: Added resistance in wind and waves
• Check requirements and determine valid added resistance methods

• Compute the added resistance for all valid methods
• Compute mean values from valid methods: Radded = Rwave + Rwind
• Compute total resistance:
Rtot = RT + Radded
Module 4: Propulsive efficiency
• Compute propulsive efficiency for total resistance: ⌘D
Module 5: Final power and plot
• Compute the final brake power: PB
54
Chapter 5. New Performance Prediction Method 5.2 Program Structure
5.2.2 Flowchart
Figure 5.5: Flowchart of the new power prediction method
55
5.2.3 Coordinate Systems

The direction of wind and waves is defined relative to the ship reference frame. A coordinate system is defined
in the program as follows.
Wave direction
The relative wave angle is defined within [0 ,±180 ], where 0 is head sea and ±180 is following sea. It is
equally defined for port and starboard side, see Figure 5.6.
Figure 5.6: Relative wave angle in the ship reference frame
Wind direction
The relative wind angle is defined within [0 ,360 ], where 0 and 360 is head wind, and 180 is following
wind. Figure 5.7 illustrates the system.
Figure 5.7: Relative wind angle in the ship reference frame
5.3 Verification of Methods

An important part of obtaining reliable power predictions is to ensure the methods and calculation procedures
are implemented correctly in the MATLAB program. This may be verified by reproducing the results to inde-
pendent calculations. As most of the procedures are well known, several calculation examples exist in the
literature. This section briefly outlines the applied verification process for the program.
An example presented in Birk (2019) is applied to verify the calm water resistance methods, i.e. Hollenbach,
Holtrop-Mennen and Guldhammer-Harvald. The example presents the power prediction of a 1000 TEU con-
tainer ship for all three methods. The container ship parameters are given as input to the MATLAB program
in order to reproduce the results from the example. As Birk (2019) applies Hollenbach and Holtrop-Mennen
as originally presented by the respective authors, these versions are also selected for the comparisons. The
program successfully reproduces the results with a deviation of 0-2% for the three different methods. The
deviations are assumed to be a result of the parameter estimates. This indicates the methods are imple-
mented correctly. However, as presented in Chapter 3, some suggested modifications to Hollenbach and
Holtrop-Mennen are implemented in the program. In order to investigate the reliability of these improvements,
56
Chapter 5. New Performance Prediction Method 5.3 Verification of Methods
a validation process with ship trial data must be conducted.
For the verification of STA-1 and STA-2 a report by MARIN (2006) is applied. The report presents estimates
for the added resistance in waves calculated by the respective methods for several ships. A 180 m product
tanker is selected for reproducing the results in MATLAB. It gives a deviation of less than 1% for the two meth-
ods. A calculation example in the ISO 15016 standard is applied for STA-JIP wind which gives a deviation
of 3%. The latter deviation is assumed to be due to the simplification of the relative wind velocity vector, as
discussed in Section 5.1.2.
No example calculations with sufficient level of details are found for the Townsin-Kwon method or the simplified
Wageningen B-series method by the author. However, the methods are computationally simple, and it is
therefore considered reasonable to verify the procedures with hand calculations. The product tanker from
Marin report is applied for Townsin & Kwon, and a model test report is applied for Wageningen B. The hand
calculations correspond to the program computations with 0 % deviation in both cases.
57
58
Chapter 6
Results and Validation
The goal of this chapter is to assess how well the new model predicts the ship power and whether the results
are improved from the current MariTEAM model. This include evaluating the performance of each method
implemented in the model as well as the parameter estimates. In order to investigate the performance, a case
study is conducted for seven ships of different type and size. The accuracy of the parameter estimates are
computed, and the predicted power for each ship is compared to the actual ship power at sea or from model
tests. Power predictions are performed both for the new model and the current MariTEAM model to compare
the results. The following section outlines the case study and the corresponding validation process.
6.1 Case Study

The objective of the case study is to test the new method on a range of ships and assess the performance.
Since the method is developed for fleet-wide calculations, seven vessels of various type and size are included
in the case study. Ideally, the case data should be representative of the whole fleet and include more than
seven ships. However, the selected cases comprise validation data which is crucial in the power prediction
performance assessment. The case vessel data is further outlined in the following.
6.2 Case Vessel Data

The case vessel data comprise detailed hull characteristics, model test reports, sea trial reports and in-
service data. Since Sea-web only provides displacement and draught on maximum load line, Sea-web data
is included for two of the ships to investigate the effect of varying loading conditions on the power prediction
performance. Table 6.1 gives an overview of the data available for each case vessel. The validation process
is described in detail in Section 6.4
Table 6.1: Overview of case vessel data in the case study
Case Ship type Model test data Sea trial data Sea-web data In-service data
(design load) (heavy ballast) (maximum load) (in-service load)
1 Cargo ship x x x
2 Container (13,000 TEU) x x x
3 Vehicles carrier x
4 Wellboat x
5 Chemical tanker x
6 Container (3,500 TEU) x
7 Bulk carrier x
Model test reports include detailed information on ship resistance and propeller performance for various load-
ing conditions and large speed ranges. All the available model test reports in this case study present full-scale
powering predictions for calm water conditions. The data is therefore applied to validate the predicted power,
59
Chapter 6. Results and Validation
resistance, and propulsion factors in calm water.
Sea trial reports typically contain fewer details about hull and propeller performance, relative to model test
reports. The two available reports in this case study comprise the power measured on heavy ballast loading
over a small speed range. Further, the power is corrected for environmental effects and presented for calm
water conditions. The reports are applied to validate the calm water power prediction for the two cases. How-
ever, the data is scarce, and the ballast loading condition deviates from the design loading assumed in the
new method.
In addition to the calm water data available, two of the cases also comprise one month of in-service data.
The measured in-service power is valuable for validating the power prediction in a seaway, including the
performance in weather. The measured power is related to the ship’s position in time and space, and for a
given speed. The ship route is then applied to connect hindcast weather data and identifies the weather state
for the ship. This is further described in the following section.
6.3 Weather Data

In-service measurements of voyage data are available for the cargo ship (Case 1) and the 13,000 TEU con-
tainer ship (Case 2). The respective routes are plotted in Figures 6.1a and 6.1b.
(a) Case 1: Cargo ship (b) Case 2: Container ship (13,000 TEU)
Figure 6.1: In-service routes for the recorded voyage data
Before the in-service measurements are analysed, the datasets are cleaned, and outliers are removed. The
in-service data for Case 1 and Case 2 are processed by Gupta et al. (2019) and Kim (2020), respectively.
This processing includes removing transient conditions and connecting hindcast weather data to the voyage.
As presented in Section 4.1.5, the temporal solution of wave and wind parameters from ECMWF (2014)
are respectively 6 hours and 3 hours. The parameters are therefore interpolated in time to fit the in-service
measurements. For in-shore areas the environmental parameters are typically zero and these parts of the
route are neglected in the validation analysis. Finally, the weather data for wind and waves are transformed
into the ship reference frame. Table 6.2 outlines the weather data available for Case 1 and Case 2.
Table 6.2: In-service weather data variables
Parameter Symbol Unit

Significant wave height Hs [m]
Mean wave period Tw [s]
Relative mean wave direction wave [ ]
True wind speed U [m/s]
Relative wind speed Urel [m/s]
Relative wind direction wind [ ]
In addition to the measured power, the in-service data comprise the actual fore and aft draught of the ship.
For Case 1, the actual displacement is also obtained by Gupta et al. (2019), using a 3D model of the ship.
60
Chapter 6. Results and Validation 6.4 Validation Methods
6.4 Validation Methods

The validation process consists of two parts. The first part is validating the calm water power requirements
without including the effects of wind and waves. This is done by comparing the calculations to model test
results or sea-trial measurements. The second part of the validation is to include the effects of wind and
waves and compare the power predictions to in-service data. Both parts are conducted both with the new
model and with the current MariTEAM model to compare the performance. Results for the new model is
presented first in Sections 6.5 - 6.7, followed by a comparison to the current MariTEAM model in Section 6.8.
6.5 Validation of Calm-Water Power Prediction

The calm water power prediction is computed as the mean of the valid methods (Guldhammer-Harvald,
Holtrop-Mennen or Hollenbach). Results for all three methods are, however, included in the validation analy-
sis. In cases where only one method is applicable, the mean is still calculated for all methods and added to
the results. The objective of this is to assess how well the calm water resistance methods and the propulsion
method perform. Mean and standard deviations are computed for the predicted power in each case.
As presented previously, the new method applies a number of parameter estimates for missing input param-
eters in Sea-web. These parameters are, however, known for the seven validation cases. The calm water
validation is therefore first conducted for a complete set of exact input parameters, without utilising the pa-
rameter estimates. This way, the implemented methods are evaluated without any influence of error from
the parameter estimates. Then, the computations are performed, including the estimated parameters to as-
sess the performance of the parameter estimates. Mean and standard deviations are computed for these
calculations as well, although presented in Section 6.8 together with the current MariTEAM model.
6.5.1 Case 1: Cargo Ship

Case 1 is a cargo ship with main particulars as listed in Table 6.3. The validation data comprise full-scale
power predictions at design draught from a model test report. Calculated results are presented in the following,
both with the complete set of exact input parameters, and when the parameter estimates are included.
Table 6.3: Main particulars of Case 1: Cargo ship at design draught
Parameter Symbol Value Unit

Length between perpendiculars Lpp 194.00 [m]
Breadth B 32.26 [m]
Draught T 12.00 [m]
Volume displacement 5 59,691.00 [m3 ]
Block coefficient CB 0.795 [ ]
Propeller diameter DP 7.00 [m]
No. of blades Z 4 [ ]
No. of propellers - 1 [ ]
Exact input parameters
Figure 6.2 presents the delivered power PD , computed by each calm water method and by the model test. All
calm water methods are applicable in this case, and the Hollenbach minimum estimate is plotted as a stapled
line as it is not included in the calculated mean.
61
Figure 6.2: Validation of power prediction for Case 1 with design loading and exact input parameters. Model test results
and calculated results.
Table 6.4 presents the deviation between the calculated results and the model test, expressed in terms of the
mean deviation plus the standard deviation. Hollenbach’s mean estimate is the most accurate in this case,
with a mean deviation of 4% and standard deviation of 5%. While all methods overestimate the power, Hollen-
bach’s minimum estimate underestimates the power by 10% on average. Guldhammer-Harvald significantly
overpredicts the power, on average by 36%. A reason for this may be that cargo ships are not included in the
modernised Guldhammer-Harvald by Kristensen et al. (2017), so the calculations are computed without the
bulb correction. Similarly, for Holtrop-Mennen, a bulb correction is not included in the method, as discussed
in Section 3.1.2.
Table 6.4: Deviation between calculated power and model test power for Case 1 (exact input parameters)
Method Mean deviation Std. deviation Applicable

Hollenbach (w/ mean resistance) 4% 5% yes
Holtrop-Mennen 19% 5% yes
Guldhammer-Harvald 36% 4% yes
Calculated mean 19% 5% yes
Hollenbach (w/ minimum resistance)* -10% 5% yes
* Not included in the calculated mean
Estimated input parameters
Figure 6.3 presents the computed results including the parameter estimates. Compared to the results with
exact input, there is little difference in the predicted power. It is slightly lower for all the computed cases,
although the trend is the same.
62
Chapter 6. Results and Validation 6.5 Validation of Calm-Water Power Prediction
Figure 6.3: Validation of power prediction for Case 1 with design loading and estimated parameters. Model test results
Table 6.5 gives the deviation between the parameter estimates and the exact input. The transverse projected
area deviates significantly by -44.7%, which gives an underpredicted air resistance. This may be the reason
for the slightly lower power predicted.
Table 6.5: Deviation of estimated parameters for Case 1 with design loading condition
Parameter Symbol Unit Ship Estimated Deviation

Length of waterline LW L [m] 198.30 197.88 -0.2 %
Draught fore TF [m] 12.00 12.00 0.0 %
Draught aft TA [m] 12.00 12.00 0.0 %
Block coefficient CB [ ] 0.795 0.795 0.0 %
Wetted surface area S 2
[m ] 9,496.00 9,155.15 -3.6%
Prismatic coefficient CP [ ] 0.798 0.800 0.2%
Transverse projected area AV T 2
[m ] 700.00 387.12 -44.7%
Propeller diameter Dp [m] 7.00 6.68 -4.6%
6.5.2 Case 2: Container Ship (13,000 TEU)

Case 2 is a 13,000 TEU container ship with main particulars as listed in Table 6.6. The validation data
comprise calm water power predictions at heavy ballast loading from a sea trial report. This loading condition
gives a sea trial draught of 7.25 m, while the design draught is 14.50 m. Further, the propeller is surface
piercing in the sea trial, as seen in Table 6.6. An underlying assumption for the new model is design draught
and submerged propeller, so the ballast loading condition is expected to give deviations in the predictions.
Calculated results are presented in the following, both with the complete set of exact input parameters, and
when the parameter estimates are included.
63
Table 6.6: Main particulars of Case 2: Container Ship (13,000 TEU) at ballast draught

Breadth B 48.20 [m]
Draught T 7.25 [m]
Figure 6.4 presents the delivered power PD , computed by each calm water method and by the sea trial. Only
Holtrop-Mennen is applicable in this case, and the other computations are plotted in stapled lines. The small
draught and displacement give too large ratios for (L/51/3 ), (B/T ) and (Dp /T ), for Guldhammer-Harvald
and Hollenbach.
Figure 6.4: Validation of power prediction for Case 2 with heavy ballast loading and exact parameters. Sea trial results
Table 6.7 presents the deviation between the calculated results and the sea trial, expressed in terms of the
mean deviation plus the standard deviation. Holtrop-Mennen’s estimate is on average the most accurate
with a mean deviation of 0%, although with a relatively large standard deviation of 9%. The other, non-valid
methods, underpredict the delivered power. In this case, the modernised Guldhammer-Harvald is applicable
and the results show that Guldhammer-Harvald’s and Hollenbach’s mean power predictions are similar.
64
Table 6.7: Deviation between calculated power and sea trial power for Case 2 (exact input parameters)

Hollenbach (w/ mean resistance) -6% 6% no
Guldhammer-Harvald -6% 7% no
Calculated mean -6% 7% no
Hollenbach (w/ minimum resistance)* -17% 6% no
exact input, there is a large difference in the predicted power. As Table 6.8 displays, the estimated propeller
diameter deviates -50.5 % the program, due to the surface piercing condition. This result in a significantly
underpredicted propeller efficiency of about 30% for all the methods. As a result, the delivered power is
significantly overpredicted. In addition, the draught fore and aft deviate largely due to the trim. The transverse
projected area is also overpredicted by 20.1%, which overpredicts the resistance.
Figure 6.5: Validation of power prediction for Case 2 with heavy ballast loading and estimated parameters. Sea trial
results and calculated results.
65
Table 6.8: Deviation of estimated parameters for Case 2 with heavy ballast loading condition

Draught fore TF [m] 4.90 7.25 48.0 %
Draught aft TA [m] 9.60 7.25 -24.5 %
Block coefficient CB [ ] 0.582 0.586 0.6%
Wetted surface area S [m2 ] 14,216.20 14,681.89 3.3%
Midship section coefficient CM [ ] 0.975 0.977 0.2%
Transverse projected area AV T [m2 ] 1,950.00 2,341.22 20.1%
Corrected propeller diameter
Figure 6.6 shows the results when the propeller diameter is corrected. As can be seen, the estimate is
significantly improved. This indicate that the estimated propeller diameter affects the predicted power more
than the other parameter estimates in this case. Note that the range on the y-axis is different in the plots.
Figure 6.6: Validation of power prediction for Case 2 with heavy ballast loading and estimated parameters, with corrected
propeller diameter. Sea trial results and calculated results.
6.5.3 Case 3: Vehicles Carrier

Case 3 is a vehicles carrier with main particulars as listed in Table 6.9. The validation data comprise full-
scale power predictions at design draught from a model test report. This vessel is twin-screw, therefore,
the underlying assumption in the method of single screw vessels is expected to give some deviations in
the predictions. Calculated results are presented in the following, both with the complete set of exact input
parameters, and when the parameter estimates are included.
66
Table 6.9: Main particulars of Case 3: Vehicles carrier at design draught

Breadth B 22.70 [m]
Draught T 6.50 [m]
Figure 6.7 presents the delivered power PD , computed by each calm water method and by the model test.
Holtrop-Mennen and Hollenbach are applicable in this case, as well as the mean of the two. Guldhamer-
Harvald is not applicable due to a too large slenderness ratio (L/51/3 ).
Figure 6.7: Validation of power prediction for Case 3 with design loading and exact parameters. Model test results and
calculated results.
mean deviation plus the standard deviation. All the applicable methods underestimate the power, whereas
Holtrop-Mennen’s estimate is the most accurate. The applicable froude number range for Guldhammer-
Harvald and Hollenbach is exceeded in this case.
67

Hollenbach (w/ mean resistance) -9% 3% yes
Holtrop-Mennen -7% 1% yes
Guldhammer-Harvald 5% 9% no
Calculated mean** -8% 1% yes
** Mean of Hollenbach and Holtrop-Mennen
exact input, the power is even more underpredicted. As Table 6.11 presents, the estimated wet surface area
deviates by -15.4%, and the transverse projected area deviates by -19.6%. These deviations result in an
underpredicted resistance, and further, an underpredicted power.
68
Table 6.11: Deviation of estimated parameters for Case 3

Length of waterline LW L [m] 134.60 133.93 -0.5 %
Draught fore TF [m] 6.50 6.50 0.0 %
Draught aft TA [m] 6.50 6.50 0.0 %
Block coefficient CB [ ] 0.573 0.572 -0.1 %
[m ] 3826.45 3,235.50 -15.4%
Prismatic coefficient CP [ ] 0.589 0.586 -0.4%
[m ] 692.00 556.20 -19.6%
Propeller diameter Dp [m] 4.50 4.55 1.1%
6.5.4 Case 4: Wellboat

Case 4 is a wellboat with main particulars as listed in Table 6.12. The validation data comprise full-scale
power predictions at design draught from a model test report. In contrast to the other vessels included in the
case study, the wellboat is not a conventional ship in the fleet. As most of the empirical calculation methods
are based on conventional hull forms, this may give some deviations in the predictions. Calculated results
are presented in the following, both with the complete set of exact input parameters, and when the parameter
estimates are included.
Table 6.12: Main particulars of Case 4: Wellboat at design draught

Breadth B 16.00 [m]
Draught T 6.40 [m]
Holtrop-Mennen is the only valid method for this case, and the other power predictions are plotted in sta-
pled lines. Guldhammer-Harvald and Hollenbach is not applicable due to a too small slenderness ratio. As
illustrated in the figure, Hollenbach’s predictions are further limited by the froude number range.
69
calculated results.
Table 6.13 presents the deviation between the calculated results and the model test, expressed in terms of
the mean deviation plus the standard deviation. Of the calm water methods, Hollenbach’s mean prediction
is the most accurate, even though it is not applicable. Guldhammer-Harvald overpredicts the power, while
Holtrop-Mennen underpredicts the power. As a result, the calculated mean is more accurate than the two
latter methods.

Calculated mean 3% 9% no
exact input, all the power predictions are somewhat reduced. As a result, the applicable prediction by Holtrop-
Mennen deviates more. As displayed in Table 6.14, the prediction of the wet surface deviates -7.3%, which
underpredicts the resistance. Further, the propeller diameter is overestimated by 35.8%. As a result the
propulsive efficiency is overpredicted. Despite this, the final predicted power is still underestimated due to the
resistance and the deviation increase for higher ship speed.
70

Length of waterline LW L [m] 75.29 74.87 -0.6%
Draught fore TF [m] 6.40 6.40 0.0%
Draught aft TA [m] 6.40 6.40 0.0%
[m ] 1,845.58 1,711.56 -7.3%
Midship section coefficient CM [ ] 0.996 0.989 -0.6%
[m ] 210.00 222.40 5.9%
6.5.5 Case 5: Chemical Tanker

Case 5 is a chemical tanker with main particulars as listed in Table 6.15. The validation data comprise
full-scale power predictions at design draught from a model test report. Calculated results are presented in
the following, both with the complete set of exact input parameters, and when the parameter estimates are
included.
71
Table 6.15: Main particulars of Case 5: Chemical tanker at design draught

Breadth B 23.50 [m]
Draught T 9.30 [m]
Figure 6.11 presents the brake power PB , computed by each calm water method and by the model test. All
calm water methods are applicable in this case, and the Hollenbach minimum estimate is plotted as a stapled
line as it is not included in the calculated mean.
calculated results.
Table 6.16 presents the deviation between the calculated results and the model test, expressed in terms
of the mean deviation plus the standard deviation. Hollenbach’s mean estimate is the most accurate in
this case, with a mean deviation of 6% and standard deviation of 2%. While all methods overestimate the
power, Hollenbach’s minimum estimate underestimates the power by -7% on average. Guldhammer-Harvald
significantly overpredicts the power, on average by 26%. Similar to the cargo ship in case 1, the reason for
this may be that chemical tankers are not included in the modernised Guldhammer-Harvald by Kristensen
et al. (2017), so the calculations are computed without the bulb correction.
72

exact input, there is little difference in the predicted power. In this case, the accuracy of Hollenbach and
Holtrop-Mennen’s predictions increases slightly. This may be due to the underestimated wetted surface area
and transverse projected area, as presented in Table 6.17. Since both methods overestimate the power with
exact input parameters, the underpredicted parameters reduce the calculated resistance and improve the
estimate.
73

Length of waterline LW L [m] 145.84 144.84 -0.7%
Draught aft TA [m] 9.30 9.30 0.0%
[m ] 5,156.80 4,910.44 -4.8%
Midship section coefficient CM [ ] 0.995 0.991 -0.3%
[m ] 460.00 131.60 -71.4%
6.5.6 Case 6: Container ship (3,500 TEU)

Case 6 is a 3,500 TEU container ship with main particulars as listed in Table 6.18. The validation data
comprise full-scale power predictions at design draught from a model test report. Calculated results are
presented in the following, both with the complete set of exact input parameters, and when the parameter
estimates are included.
Table 6.18: Main particulars of Case 6: Container ship (3,500 TEU) at design draught

Breadth B 32.20 [m]
Draught T 11.00 [m]
Only Holtrop-Mennen is applicable in this case, and the other computations are plotted in stapled lines. The
slenderness ratio and the length-breadth ratio are too large for Guldhammer-Harvald and Hollenbach.
74
calculated results.
mean deviation plus the standard deviation. Holtrop-Mennen’s estimate is on average the most accurate with
a mean deviation of 2%, and a standard deviation of 6%. The other, non-valid methods, underestimate the
delivered power. As for the other container ship in Case 2, the modernised Guldhammer-Harvald is applicable,
and the results show that Guldhammer-Harvald’s and Hollenbach’s minimum power predictions are similar.

exact input, there is little difference in the predicted power, and the trend is the same. Table 6.20 gives the de-
viation between the parameter estimates and the exact input. The transverse projected area is overestimated
by 48.5%, and the propeller diameter is underestimated by 11%. Although these deviations are significant,
the predicted power is not significantly affected.
75

Length of waterline LW L [m] 226.66 235.33 3.8%
Draught fore TF [m] 11.00 11.00 0.0%
Draught aft TA [m] 11.00 11.00 0.0%
[m ] 8,979.13 9,163.47 2.1%
Prismatic coefficient CP [ ] 0.589 0.586 -0.4%
[m ] 850.00 1,262.24 48.5%
6.5.7 Case 7: Bulk Carrier

Case 7 is a bulk carrier with main particulars as listed in Table 6.21. The validation data comprise calm water
power predictions at heavy ballast loading from a sea trial report. This loading condition gives a sea trial
draught of 8.15 m, while the design draught is 12.2 m. As discussed for the sea trial validation of Case 2,
the underlying assumption for the new model is design draught, so the ballast loading condition is expected
to give deviations in the predictions. Further, the sea trial report only provide three data points over a small
speed range. Calculated results are presented in the following, both with the complete set of exact input
parameters, and when the parameter estimates are included.
76
Table 6.21: Main particulars of Case 7: Bulk Carrier at ballast draught

Breadth B 32.27 [m]
Draught T 8.15 [m]
Figure 6.15 presents the brake power PB , computed by each calm water method and by the sea trial. Only
Holtrop-Mennen is applicable in this case, and the other computations are plotted in stapled lines. Similar to
Case 2, the small draught and displacement result in a too large slenderness ratio for Guldhammer-Harvald
and Hollenbach.
Figure 6.15: Validation of power prediction for Case 7 with heavy ballast loading and exact parameters. Sea trial results
Table 6.22 presents the deviation between the calculated results and the sea trial, expressed in terms of the
mean deviation plus the standard deviation. Holtrop-Mennen’s estimate is the only applicable and also the
most accurate, with a mean deviation of 2%, and standard deviation of 2%. Hollenbach’s mean estimate is
slightly underestimated, on average by 3%. In this case, the modernised Guldhammer-Harvald is applied,
and the results show that Guldhammer-Harvald’s and Hollenbach’s mean power predictions are similar. The
minimum prediction by Hollenbach represents the most significant deviation of -13 %.
77
Table 6.22: Deviation between calculated power and sea trial power for Case 7 (exact input parameters)

Hollenbach (w/ minimum resistance) -13% 2% no
exact input, there is a large difference in the predicted power. Note that the range on the y-axis is different in
the plots. The tendency of the results is similar to the sea trial results from Case 2. As Table 6.23 displays,
the estimated propeller diameter deviates -37.2 %, which is expected to result in an underestimated propeller
efficiency. This is, however, not verified, as the sea trial report only provides brake power data. In addition,
the draught fore and aft deviate largely due to the trim. The transverse projected area is also overpredicted by
41.2%, which overestimates the resistance. An underestimated propulsive efficiency, and an overestimated
resistance results in a significantly overpredicted power.
Figure 6.16: Validation of power prediction for Case 7 with heavy ballast loading and estimated parameters. Sea trial
results and calculated results.
Table 6.23 lists less parameters than for the other cases. The number of estimated parameters in the calcu-
lations are naturally the same for all the cases, however, the sea trial report only provides the exact values for
the parameters included below.
78
Chapter 6. Results and Validation 6.6 Validation of In-service Power Prediction

Draught aft TA [m] 8.96 8.15 -9.0%
[m ] 384.93 543.58 41.2%
Corrected propeller diameter
Figure 6.17 shows the results when the propeller diameter is corrected. As can be seen, the estimate is signif-
icantly improved. This indicates that the estimated propeller diameter affects the predicted power more than
the other parameter estimates in this case. However, the power is still somewhat overestimated compared to
the exact input results for each method.
Figure 6.17: Validation of power prediction for Case 7 with heavy ballast loading, estimated parameters but corrected
propeller diameter. Sea trial results and calculated results.
6.6 Validation of In-service Power Prediction

The second part of the validation includes the effects of wind and waves and compares the power predictions
to the in-service data of Case 1 and Case 2. In addition to the computed mean added resistance, the per-
formance of each added resistance method is assessed in each case. Further, the effect of varying loading
conditions on the power prediction is analysed. Each case is analysed for both the in-service draught and the
maximum draught provided by Sea-web, to investigate the difference. The root-mean-square error (RMSE)
between the measured power and the calculated power is applied to assess the performance of the predic-
tions.
For both the in-service cases, it is assumed that the power produced by the main engine corresponds to the
propulsion power and that the other generators provide auxiliary power. Further, sea passage is the only state
considered, while manoeuvring or anchor/waiting is disregarded.
79

The voyage data of Case 1 include one month of measurements of ship speed, draught and power produced
by the engines. In addition, the in-service displacement is made available by Gupta et al. (2019). The mean
in-service draught and displacement is applied in the following calculations, together with the weather data
variables from the voyage. Table 6.24 presents an overview of the weather characteristics. As displayed, the
weather range from calm sea (BN = 0) to strong breeze and large waves (BN = 6). Both the wind and
wave heading vary from head-to to following.
Table 6.24: In-service weather data range for Case 1
Parameter Symbol Range [min, max] Unit

Significant wave height Hs [0.08,3.64] [m]
Mean wave period Tw [0.37,12.35] [s]
Relative mean wave direction wave [1.26, 180.0] [ ]
True wind speed U [0.10, 11.75] [m/s]
Beauforts number BN [0,6] [ ]
Relative wind speed Urel [0.45, 13.79] [m/s]
Relative wind direction wind [1.07, 358.90] [ ]
Figure 6.18 presents the measured brake power (in dark blue), the calculated power by the new method
(middle blue), and the computed calm water power (light blue). These results are calculated by applying
the mean added resistance estimate. The predicted power follows the trend of the measured power over
the speed range. There is, however, measured a constant in-service power for a range of speed. This is
observed by the amount of in-service scatter points, forming the horizontal shape in the plot, which deviates
from the typical cubic relation between power and speed. A reason for this may be uncertainties related to
the on-board measurements of the ship speed. Also, the predicted power appears slightly overestimated in
the lower speed range. This corresponds to the results from the calm water analysis of the model test.
Figure 6.18: Validation of power prediction for Case 1 with in-service loading. In-service measurements and calculated
results.
Added resistance module performance
Several combinations of methods to predict the added resistance in waves and wind are investigated. RMSE
is computed for each combination and for the mean of all methods. As presented in Table 6.25, the RMSE of
80
the computed mean is 572.02 kW. Applying Kwon alone increases the error by 7%, to 610.83 kW. The most
accurate prediction is obtained by applying STA-1 for wave resistance and Blendermann for wind resistance,
improving the accuracy by 5% relative to the mean prediction. Figures 6.19a and 6.19b illustrate the most
accurate and the least accurate prediction, respectively. There is a significant difference between the two,
however, the other variations of RMSE are relatively small for the various combinations.
Table 6.25: RMSE for combinations of added resistance methods applied to Case 1 with in-service loading condition
Module combination RMSE [kW] Difference from ’mean of all methods’

Mean of all methods 572.02 -
Townsin & Kwon 610.83 7%
Mean wind + mean(STA-1, STA-2) 551.20 -4%
Mean wind + STA-1 550.24 -4%
Mean wind + STA-2 552.31 -3%
Mean wave + STAJIP wind 559.97 -2%
Mean wave + Blendermann 543.97 -5%
Best combination: STA-1 + Blendermann 543.03 -5%
(a) STA-1 (wave) + Blendermann (wind) (b) Townsin & Kwon (wind and waves combined)
Figure 6.19: The added resistance modules with highest accuracy (a) and lowest accuracy (b) in terms of RMSE
Effect of loading condition
For fleet-wide calculations, the exact in-service draught is unavailable. Sea-web only provides maximum
draught, which is expected to affect the power predictions significantly. For Case 1, the the mean in-service
draught is 7.91 m, and the maximum draught provided by Sea-web is 12.64 m. The maximum draught is
applied to the calculations and the result is presented in Figure 6.20. The predicted power is overestimated,
both relative to the in-service measurements and to the prediction on in-service draught. Table 6.26 presents
the RMSE of the prediction, which is increased 197% relative to the in-service draught prediction. In this
case, the loading condition affects the predicted power significantly more than the choice of added resistance
method.
81
Figure 6.20: Validation of power prediction for Case 1 with maximum loading from Sea-web. In-service measurements
Table 6.26: Deviation of predicted power for in-service loading condition and for maximum loading condition
Loading condition Draught Added resistance calculation method RMSE [kW]

In-service loading 7.91 m Mean of all methods 572.02
Maximum loading 12.64 m Mean of all methods 1699.30
6.6.2 Case 2: Container Ship

The voyage data of Case 2 include one month of measurements of ship speed, draught and power produced
by the engines. In this case, the in-service displacement (and dwt) is not known. Therefore, the displacement
is retrieved from Sea-web. The mean in-service draught is applied in the following calculations, together with
the weather data variables from the voyage. Table 6.27 presents an overview of the weather characteristics.
As displayed, the weather range from light conditions (BN = 1) to moderately strong conditions (BN = 8).
It is observed that the maximum wave height does not correspond to typical values for BN = 8. In this case,
the wind is measured onboard the ship, and may, therefore, include some effects of wind gusts. Both the wind
and wave heading vary from head-to to following.
Table 6.27: In-service weather data range for Case 2
Parameter Symbol Range [min, max] Unit

Significant wave height Hs [0.00,2.34] [m]
Mean wave period Tw [0.10,14.00] [s]
Relative mean wave direction wave [0.30, 180.0] [ ]
True wind speed U [0.67,17.56] [m/s]
Beauforts number BN [1,8] [ ]
Relative wind speed Urel [0.15,24.41] [m/s]
Relative wind direction wind [0.00,360.00] [ ]
Figure 6.21 presents the measured brake power and the calculated power by the program. These results
are calculated by applying the mean added resistance estimate. The predicted power is somewhat underes-
timated, but follows the trend of the measured power over the speed range. Even though the exact draught
82
is applied, the displacement is retrieved from Sea-web and is therefore probably overestimated. This is ex-
pected to overpredict the power, hence is not the reason for the underestimated result. A different reason may
be that the container ship has significant growth of fouling or hull roughness contributing to the resistance.
RMSE values are presented in the following.
Figure 6.21: Validation of power prediction for Case 2 with in-service loading. In-service measurements and calculated
results.
Added resistance module performance
Several combinations of methods to predict the added resistance in waves and wind are assessed. The RMSE
is computed for each combination and for the mean of all methods. As presented in Table 6.28, the RMSE
of the computed mean is 3,694.11 kW. Applying Kwon alone increases the error by 4%, to 3,827.74 kW. The
most accurate prediction is obtained by applying STA-1 for wave resistance and STAJIP for wind resistance,
improving the accuracy by 3% relative to the mean prediction. Figures 6.22a and 6.22b illustrate the most
accurate and the least accurate prediction, respectively. Similar as for Case 1, there is a significant difference
between the two, however, the other variations of RMSE are relatively small for the various combinations.
Table 6.28: RMSE for combinations of added resistance methods applied to Case 2 with in-service loading condition
Module combination RMSE [kW] Difference from ’mean of all methods’

Mean of all methods 3,694.11 -
Townsin & Kwon only 3,827.74 4%
Mean wind + mean(STA-1, STA-2) 3,615.02 -2%
Mean wind + STA-1 3,614.45 -2%
Mean wind + STA-2 3,628.23 -2%
Mean wave + STAJIP wind 3,586.54 -3%
Mean wave + Blendermann 3,650.42 -1%
Best combination: STA-1 + STAJIP wind 3,585.89 -3%
83
(a) STA-1 (wave) + STAJIP (wind) (b) Townsin & Kwon (wind and waves combined)
Figure 6.22: The added resistance modules with highest accuracy (a) and lowest accuracy (b) in terms of RMSE
Effect of loading condition
The effect of varying loading conditions is not properly captured in this case, since the in-service displacement
is unavailable. Even though the in-service draught is available, important parameters depend on the displace-
ment such as the block coefficient and the wetted surface area. Therefore, a varying loading condition is
not completely captured by only changing the draught. A power prediction is still computed for the maximum
draught, presented in Figure 6.23.
Figure 6.23: Validation of power prediction for Case 2 with maximum loading from sea-web. In-service measurements
In this case, the computed RMSE for the maximum loading is improved, as Table 6.29 displays. The reason for
this may be that the overestimated draught contribute to increase the power which is initially underestimated,
as discussed above. These results are, therefore, not considered to be significant.
84
Chapter 6. Results and Validation 6.7 Summary of Results and Validation
Table 6.29: Deviation of predicted power for in-service loading condition and for maximum loading condition
Loading condition Draught Added resistance calculation method RMSE [kW]

In-service loading 13.12 m Mean of all methods 3,694.11
Maximum loading 15.52 m Mean of all methods 3,367.78
6.7 Summary of Results and Validation

This section is dedicated to summarising the results from the validation of the new model. First, the results
from the calm water power predictions are summarised, followed by the in-service predictions and finally, the
parameter estimates. The objective is to give an overview of the performance of the program.
6.7.1 Performance of the Calm Water Resistance Methods

In the validation of the calm water power predictions, all methods are included for each case. However, not all
methods are applicable, and Table 6.30 summarises the applicability. Holtrop-Mennen applies to all the cases
(indicated in green), while Hollenbach only applies to three ships, and Guldhammer-Harvald to two. As most
of the ships analysed are conventional vessel types, all methods were expected to apply. However, Case 2
(container ship), and Case 7 (bulk carrier) have ballast loading conditions which are outside the applicable
area of Hollenbach. Further, the wellboat in Case 4 does not represent a conventional hull form.
Table 6.30: Applicable methods for Case 1 - Case 7
Case Ship type Hollenbach Holtrop-Mennen Guldhammer-Harvald

1 Cargo ship
2 Container (13,000 TEU)
3 Vehicles carrier
4 Wellboat
5 Chemical tanker
6 Container (3,500 TEU)
7 Bulk carrier
In order to assess the accuracy of the calm water resistance methods, the results for exact input are eval-
uated, excluding any error of the parameter estimates. Table 6.31 presents the mean deviation and mean
standard deviation for all cases. The mean deviation for all cases indicates whether the method on average
underestimates or overestimates the results. Cancellation effects may not be captured in the final mean; how-
ever, the variability is expressed in terms of the standard deviation. The presented results include deviation of
predicted calm water resistance, propulsive efficiency and final power. All methods are included, even in the
cases where they are not applicable.
Hollenbach’s mean resistance estimate is on average the most accurate with only 1% deviation, even though
the method is not applicable for all the ships. Holtrop-Mennen follows with a mean deviation of 5%. Guldhammer-
Harvald overestimate the resistance on average by 11%. This may be explained by the limited applicability
of the modernised method, only applied to 3 out of 7 cases. Hollenbach’s minimum estimate underestimate
the resistance, on average by 8%. These large variations are cancelled out for the calculated mean, which on
average deviates only 5%.
The propulsive efficiency is predicted within an accuracy of 3% for all the cases. This indicate that the
estimated thrust and wake factors from the calm water methods are reasonably reliable and that the simplified
Wageningen-B series perform well. As highlighted in the Table, Holtrop-Mennen and Hollenbach’s estimates
for the final predicted power are evidently the most accurate in the comparison.
85
Table 6.31: Mean deviation and mean std. deviation for the calculated resistance, propulsive efficiency and power.
Results include all cases.
Resistance Propulsive efficiency Power

Mean Std. Mean Std. Mean Std.
deviation deviation deviation deviation deviation deviation
Hollenbach (w/ mean resistance) 1% 4% 2% 2% -3% 5%
Holtrop-Mennen 5% 5% 3% 2% 2% 6%
Guldhammer-Harvald 11% 6% 3% 2% 11% 6%
Calculated mean 5% 5% 2% 2% 2% 5%
Hollenbach (w/ min. resistance) -8% 3% 3% 2% -16% 4%
The results in Table 6.32 only include the applicable cases. Since Holtrop-Mennen is valid for all cases, these
results are equal as for Table 6.31. However, for Hollenbach, both the mean and the minimum predictions
are improved. Guldhammer-Harvald’s average deviations increase significantly. The reason may be that
Guldhammer-Harvald is only valid for two cases, Case 1 (cargo ship) and Case 5 (chemical tanker), and both
cases are only valid to analyse with the old Guldhammer-Harvald method. The modernised corrections by
Kristensen et al. (2017) are therefore not applied in these cases.
Results only include cases where the respective methods are valid.

Hollenbach (w/ mean resistance) 1% 3% 0% 1% 0% 3%
Holtrop-Mennen 5% 5% 3% 2% 2% 6%
Guldhammer-Harvald 22% 4% -8% 1% 31% 5%
Calculated mean 7% 3% -1% 1% 8% 3%
Hollenbach (w/ min. resistance) -3% 3% 0% 1% -13% 3%
To summarise these results, Hollenbach’s mean estimate is the most accurate overall, followed by Holtrop-
Mennen. Both Guldhammer-Harvald and Hollenbach’s minimum power predictions deviate significantly, on
average more than 10%.
6.7.2 Performance of the In-service Prediction

The in-service power predictions for Case 1 (cargo ship) and Case 2 (13,000 TEU container ship) correspond
relatively well to the voyage measurements when the correct loading condition is applied. As demonstrated
for Case 1, the effect of varying the added resistance methods is less significant than varying the loading
condition. However, when comparing the individual methods, Townsin & Kwon underestimates the added re-
sistance relative to the other methods. The application of Townsin & Kwon alone only increases the resistance
slightly from the calm water resistance.
In Case 2, the calculated power is underestimated relative to the measured power. One reason may be the
effects of roughness or fouling on the hull, which should be investigated further in future work. Another reason
may be that the calm water power is underpredicted for the ship in this case. The latter is difficult to verify
without more information from model tests for the ship.
6.7.3 Performance of the Parameter Estimates

This section presents the deviations of the results when the parameter estimates are included in the compu-
tations. Mean deviation and standard deviation for the computed resistance, propulsive efficiency and final
power are presented. In addition, the mean deviation of each parameter estimate is calculated. The objective
86
Chapter 6. Results and Validation 6.8 Comparing the New Model to the Current MariTEAM Model
is to assess whether errors in the parameter estimates significantly influences the results.
Table 6.33 displays the mean deviation of each parameter estimate in the case study. The most significant
errors include the transverse projected area, the propeller diameter, effects of trim on the draught, and the
wet surface area.
Table 6.33: Mean deviation and mean standard deviation of the estimated parameters
Parameter Symbol Unit Mean deviation Std. deviation

Length of waterline LW L [m] 0% 2%
Draught fore TF [m] 9% 18 %
Draught aft TA [m] -5 % 9%
Block coefficient CB [ ] 0% 0%
[m ] -4 % 7%
Midship section coefficient CM [ ] 0% 1%
Prismatic coefficient CP [ ] 0% 0%
[m ] -3 % 45 %
Propeller diameter Dp [m] -8 % 29 %
While the fore and aft draught deviations are due to the assumption of zero trim, the other parameter estimates
are based on empirical formulas. The deviations indicate these formulas are not reliable for all ship types.
However, it should be noted that the propeller diameter is estimated based on the maximum draught, and
all the analysed ships have design draught or ballast draught in the case study. This is a weakness of
the calculations in the case study. However, the parameter estimate is intended to supplement Sea-web
parameters, which applies maximum draught. Table 6.34 outlines the results obtained with the estimated
parameters. All cases are included, even when the calm water method is not applicable.
Results include all cases with estimated parameters.

Hollenbach (w/ mean resistance) -3% 5% -1% 2% 2% 5%
Holtrop-Mennen 2% 5% -1% 2% 10% 6%
Hollenbach (w/ min. resistance) -16% 3% -1% 2% -12% 4%
Compared to the results with exact input parameters, the tendency is that most of the predictions are less
accurate. However, apart from Hollenbach’s minimum estimate, all the method’s resistance predictions are
still within 6% on average. Holtrop-Mennen and Guldhammer-Harvald’s power prediction deviates signifcantly
more, 10% and 17% respectively. In contrast, the deviation of Hollenbach’s mean and minimum power es-
timate is improved, to 2 % and -12% on average. The propulsive efficiency is generally underestimated by
including the parameter estimates. As demonstrated in the case study, deviations in the propeller diameter
affects the propulsive efficiency considerably.
6.8 Comparing the New Model to the Current MariTEAM Model

This section is dedicated to comparing the performance of the new model to the current MariTEAM model.
Power predictions are computed by both methods, for all the ships in the case study. Both methods include
87
their respective parameter estimates. The current MariTEAM model is implemented as outlined previously in
Section 4.2.

Case 1 is a cargo ship with main particulars as listed previously in Table 6.3. The validation data comprise
full-scale power predictions at design draught from a model test report, and one month of in-service data.
Calculated results are presented in the following, both by the new model and the current MariTEAM model.
Calm-water power prediction
Figure 6.24 presents the computed power together with the model test results. As displayed in Table 6.35, the
current MariTEAM model’s prediction overestimates the power by 22% on average, and deviates largely from
the minimum resistance estimate by Hollenbach in the new model. These two estimates were expected to be
similar, since the current MariTEAM model applies Hollenbach’s minimum resistance estimate. However, in
the new model, Hollenbach is modified as presented in Section 3.1.3 and the resistance prediction is therefore
different. Further, the propulsive efficiency is underestimated in this case by the current MariTEAM model,
contributing to an increased power.
Figure 6.24: Comparison of power predicted by the model test, new model, and the current MariTEAM model for Case 1
(parameter estimates included).
Table 6.35: Deviation of power predicted by the model test, new model, and the current MariTEAM model for Case 1

Current MariTEAM model 22% 4% yes
88
In-service power prediction
Figures 6.25a and 6.25b present the in-service power prediction by both models. In both cases the pre-
dicted power follows the trend of the in-service measurements. Similar to the calm water results, the current
MariTEAM model overpredicts the power relative to the new model. Further, Townsin & Kwon predicts less
added resistance than the methods implemented in the new model. Table 6.36 shows that the RMSE is higher
for the current MariTEAM model than the new model. Hence, the accuracy of the predicted added resistance
is improved for this case.
(a) New model (b) Current MariTEAM model
Figure 6.25: Comparison of measured in-service power to the power predicted by the new model (a) and the current
MariTEAM model (b) for Case 1
Table 6.36: Deviation of power measured in-service, predicted by the new model, and the current MariTEAM model for
Case 1.
Model Added resistance calculation method RMSE [kW]

New model Mean of all methods 572.02
Current MariTEAM model Townsin and Kwon 713.03

Case 2 is a 13,000 TEU container ship with main particulars as listed previously in Table 6.6. The validation
data comprise calm water power predictions at heavy ballast draught from a sea trial report, and one month
of in-service data. Calculated results are presented in the following, both by the new model and the current
MariTEAM model.
Calm-water power prediction
Figure 6.26 presents the computed power together with the sea trial results. Only Holtrop-Mennen is valid
to apply in this case. The results are presented without any correction of the propeller diameter, which is
underestimated by 50 % in the new model due to the surface piercing propeller. As discussed previously, this
result in a largely underestimated propeller efficiency, and correspondingly an overpredicted power.
89
Figure 6.26: Comparison of power predicted by the sea trial, new model, and the current MariTEAM model for Case 2
As Table 6.37 displays, the prediction by the current MariTEAM model is significantly more accurate than
by the new model. The current model applies a constant value for the propeller efficiency, resulting in an
improved prediction of the propulsive efficiency, which is independent of the propeller diameter.
Table 6.37: Deviation of power predicted by the sea trial, new model, and the current MariTEAM model for Case 2

Hollenbach (w/ mean resistance) 34% 9% no
Hollenbach (w/ minimum resistance)* 19% 9% no
Current MariTEAM model -9% 5% no
Calm-water resistance prediction
Due to the relatively large deviations in predicted power between the two models, the predicted resistance
is further investigated. Figure 6.27 presents the computed resistance and the sea trial resistance. In this
case, the predicted resistance by the current MariTEAM model and by Hollenbach’s minimum procedure in
the new model correspond more closely. As Table 6.38 demonstrates, Hollenbach’s mean estimate gives the
most accurate resistance prediction. The results indicate that the current MariTEAM model’s prediction of
the propulsive efficiency is better than the Wageningen-B procedure in cases where the propeller diameter is
inaccurately estimated.
90
Figure 6.27: Comparison of calm water resistance predicted by the sea trial, new model, and the current MariTEAM
model for Case 2 (parameter estimates included).
Table 6.38: Deviation of calm water resistance predicted by the sea trial, new model, and the current MariTEAM model
for Case 2 (parameter estimates included).

In-service power prediction
Figures 6.28a and 6.28b present the in-service power prediction by both models. In both cases the predicted
power follows the trend of the in-service measurements, although somewhat underestimated. Similar to Case
1, Townsin & Kwon predicts less added resistance than the methods implemented in the new model. Table
6.39 shows that the RMSE is higher for the current MariTEAM model relative to the new model.
91
(a) New model (b) Current MariTEAM model
Figure 6.28: Comparison of measured in-service power to the power predicted by the new model (a) and the current
MariTEAM model (b) for Case 2
Table 6.39: Deviation of power measured in-service, predicted by the new model, and the current MariTEAM model for
Case 2.
Model Added resistance calculation method RMSE [kW]

New model Mean of all methods 3694.11
Current MariTEAM model Townsin and Kwon 4185.94
6.8.3 Case 3: Vehicles Carrier

Case 3 is a vehicles carrier with main particulars as listed previously in Table 6.9. The validation data comprise
full-scale power predictions at design draught from a model test report. Figure 6.29 presents the computed
power by both models, together with the model test results. Only Holtrop-Mennen and Hollenbach are valid in
this case.
92
In this case, the current MariTEAM model prediction is close to the Hollenbach minimum prediction. Both
underestimate the power, on average by 32 % and 31 % as displayed in Table 6.40. As discussed in Section
6.5.3, the wet surface is underestimated by 15.4% in this case, which underestimates the resistance. Both
the current MariTEAM model and the new model applies the same parameter estimate for the wetted surface.

Calculated mean -20% 2% yes
Hollenbach (w/ minimum resistance) -31% 1% yes
Current MariTEAM model -32% 3% yes
6.8.4 Case 4: Wellboat

Case 4 is a wellboat with main particulars as listed previously in Table 6.12. The validation data comprise
full-scale power predictions at design draught from a model test report. Figure 6.30 presents the computed
power by both models, together with the model test results. Only Holtrop-Mennen is valid in this case.
In this case, all predictions show significant deviations, as presented in Table 6.41. As discussed previously,
the wellboat represents the least conventional ship type in the study. The wetted surface area is underesti-
mated in both the new and the current MariTEAN model. Note that the froude limitation range is not applied
in the current MariTEAM model.
93

Hollenbach (w/ minimum resistance) -32% 3% no
6.8.5 Case 5: Chemical Tanker

Case 5 is a chemical tanker with main particulars as listed previously in Table 6.15. The validation data
comprise full-scale power predictions at design draught from a model test report. Figure 6.31 presents the
computed power by both models, together with the model test results. All methods are applicable in this case.
In this case, the current MariTEAM model prediction is significantly more accurate than the minimum predic-
tion by Hollenbach. As displayed in Table 6.42, the average deviation is 3 %, against -15 % for Hollenbach’s
minimum estimate. The reason that these results deviate, is the underestimated propulsive efficiency in the
current MariTEAM model. The current model underestimates the resistance by 5%, and since the propulsive
efficiency also is underestimated by 9 %, the final power prediction increases.
94

Current MariTEAM model 3% 4% yes

Case 6 is a 3,500 TEU container ship with main particulars as listed previously in Table 6.18. The validation
data comprise full-scale power predictions at design draught from a model test report. Figure 6.32 presents
the computed power by both models, together with the model test results. Only Holtrop-Mennen is valid in
this case.
As Table 6.43 presents, the prediction by the current MariTEAM model is close to the minimum Hollenbach
prediction. Both predictions underestimate the power significantly, on average by 17 % and 19 % respectively.
95

6.8.7 Case 7: Bulk Carrier

Case 7 is a bulk carrier with main particulars as listed previously in Table 6.21. The validation data comprise
calm water power predictions at heavy ballast draught from a sea trial report. Calculated results are presented
in Figure 6.33, both by the new model and the current MariTEAM model. Only Holtrop-Mennen is valid
to apply in this case. The results are presented without any correction of the propeller diameter, which
is underestimated by 37.2 % in the new model. Similar to Case 2, this result in a largely underestimated
propeller efficiency, and correspondingly an overpredicted power.
Figure 6.33: Comparison of power predicted by the sea trial, new model, and the current MariTEAM model for Case 7
As Table 6.44 displays, the prediction by the current MariTEAM model is significantly more accurate than by
the new model, on average with 1 % deviation. The current MariTEAM model predicts the propulsive efficiency
with less error, as the calculation is independent of propeller diameter.
96
Table 6.44: Deviation of power predicted by the sea trial, new model, and the current MariTEAM model for Case 7

Hollenbach (w/ minimum resistance) 6% 2% no
Current MariTEAM model 1% 2% no
6.8.8 Summary of Comparison

Table 6.45 summarises the mean deviations for Case 1-7. The table include all results, even when the respec-
tive methods are not applicable. When comparing the resistance predictions, the current MariTEAM model
is close to the minimum prediction by Hollenbach. This corresponds with the expectation, since the current
MariTEAM model applies the Hollenbach minimum prediction, although not the modified version as described
in Section 3.1.3.
In the current MariTEAM model, the propulsive efficiency is computed by the empirical formula by Lindstad
et al. (2011). On average for the seven cases, the prediction deviates by -5%, which is less accurate than the
Wageningen B-series estimate with Hollenbach minimum prediction. However, Case 2 and Case 7 demon-
strate that the formula by Lindstad et al. (2011) is more accurate than the Wageningen B-series in the cases
where the propeller diameter is inaccurately estimated. The propulsive efficiency predictions in the new model
is not as reliable for surface piercing propellers or ballast draught conditions.
The final power predicted by the current MariTEAM model deviates on average by -7%, which is more ac-
curate than the prediction by Hollenbach’s minimum estimate, and less accurate than Hollenbach’s mean
prediction. In some of the cases, the current MariTEAM model benefits from underpredicting the propulsive
efficiency, since it results in an increased power when the resistance is underestimated. This cancellation
effect may improve the final power prediction, although it is not considered to increase the reliability of the
results.
Table 6.45: Deviations for the calculated resistance, propulsive efficiency and power. Results include all cases with
estimated parameters, for the new model and the current MariTEAM model.

Hollenbach (w/ mean resistance) -3% 5% -1% 2% 2% 5%
Holtrop-Mennen 2% 5% -1% 2% 10% 6%
Hollenbach (w/ min. resistance) -16% 3% -1% 2% -12% 4%
Current MariTEAM model -13% 5% -5% 4% -7% 5%
97
98
Chapter 7
Discussion
This thesis has identified, developed, and validated a ship powering performance method suitable for the
MariTEAM model. The current state of knowledge in the field of global fleet-wide power predictions has
been reviewed. In line with the literature, various empirical methods are implemented in the model. The new
method requires few input parameters and can predict the propulsion power in realistic sea states, for a wide
range of ships in the fleet.
7.1 The New Performance Prediction Method

Based on a comprehensive literature study, a wide range of empirical methods are reviewed and narrowed
down to the most relevant. These are included in the model, which is subdivided into five main modules:
• Module 1: Estimate missing input parameters

• Module 2: Calm water resistance
• Module 3: Added resistance in wind and waves
• Module 4: Propulsive efficiency
• Module 5: Final power
Three calm water resistance methods are considered to be suitable for fleet-wide calculations today, including
Hollenbach, Holtrop-Mennen, and the modernised Guldhammer-Harvald by Kristensen et al. (2017). Each
method is valid for ships with characteristics within a defined range. Module 2 returns the calm water re-
sistance as an average of the valid methods, which is based on the assumption that the three methods are
equally reliable. The work presented by Tillig (2020) and Kristensen et al. (2017) forms the basis for this
assumption. Both studies present power predictions of good accuracy by the application of Hollenbach and
the updated Guldhammer-Harvald procedure. Besides, Holtrop-Mennen is widely applied in the literature and
represents the method with the greatest applicable range. As Chapter 5 demonstrates, only 53.4% of the
fleet can be analysed by either Hollenbach or Guldhammer-Harvald, while the number of ships increases to
72.8 % if Holtrop-Mennen is included. These numbers include the applicable range defined in the original
Guldhammer-Harvald method. The updated Guldhammer-Harvald procedure by Kristensen et al. (2017) only
applies to tankers, bulk carriers and container ships, which is currently not reflected in the defined valid range.
This is considered a weakness in the model, although easy to improve in further work.
Module 3 includes Townsin & Kwon, as well as two methods to predict added resistance in waves, STAWAVE-1
and -2, and two methods to predict added resistance in wind, STAJIP wind and Blendermann. Townsin & Kwon
represents the most straightforward method and applies to all ships and sea states without restrictions, which
is suitable for fleet-wide calculations. However, the method calculates a combined added resistance in wind
and waves based on only the Beaufort number as an environmental parameter. Significant resistance contri-
butions from large superstructures or wave-induced motions are not captured; therefore, additional methods
are implemented. MARIN (2006) presents accurate predictions of added wave resistance for both STAWAVE
methods. STA-JIP wind is recommended for modern ships in ISO 15016, and Blendermann applies to an
extensive range of ship types. The average added resistance from these methods is expected to provide a
reliable prediction. However, the accuracy of the prediction depends on accurate input parameters, which
has proven to be a challenge. STA-1 requires the length of the bow to 95 % of the maximum breadth on the
99
Chapter 7. Discussion
waterline, LBW L , which is difficult to determine based on limited input. In the current work, comparison ships
are applied to estimate the value; however, a suggestion for further work is to develop a parameter estimate
based on typical hull forms within the ship types.
7.2 Validation Results

A validation study, including seven cases, has been conducted for the new model. The predicted power for
each ship is compared to the actual ship power from model tests, sea trials or in-service measurements.
Model test reports include detailed information on the ship powering performance for various loading condi-
tions, and represent the most reliable validation data source for calm water power predictions. Model test
reports are available for five out of seven cases, and sea trial reports are applied for the two remaining cases.
The sea trial reports contain fewer details about hull and propeller performance, relative to model test reports.
Also, the trials are performed on heavy ballast loading, which is an unconventional loading condition for ships
in the fleet. Therefore, model test reports serve as a significantly more solid validation data source than sea
trial reports.
7.2.1 Calm Water Powering Performance

The study evaluates the performance of each method implemented in the model as well as the parameter
estimates. In order to assess the accuracy of the calm water resistance methods, the results for exact input
are evaluated, excluding any error of the parameter estimates. The computed propulsive efficiency was on
average, predicted within 3% and with a mean standard deviation of 2% for all the cases analysed. This
indicates that the Wageningen B-series procedure is highly reliable when the propeller diameter is known.
Further, the power predictions reflect the performance of the calm water resistance prediction, since errors in
the propulsive efficiency are small. Table 7.1 summarises the mean deviation and mean standard deviation
for the calculated power in the study. The first column includes all cases, while the second only includes cases
where the particular method was applicable.
Table 7.1: Mean deviation and mean std. deviation for the calculated power (Results are computed with exact input
parameters).
All cases Valid cases

Mean Std. Mean Std. No. of valid cases
deviation deviation deviation deviation
Hollenbach (w/ mean resistance) -3% 5% 0% 3% 3
Holtrop-Mennen 2% 6% 2% 6% 7
Guldhammer-Harvald 11% 6% 31% 5% 2
Calculated mean 2% 5% 8% 3% 3
Hollenbach (w/ min. resistance) -16% 4% -13% 3% 3
Holtrop-Mennen was the only method applicable to all the cases. On average, Holtrop-Mennen’s power pre-
diction deviates 2% with a standard deviation of 6% for all the cases. Hollenbach was only applicable to three
cases, and for the particular cases, Hollenbach’s mean power prediction deviates on average by 0%, with a
standard deviation of 3%. If Hollenbach’s mean power predictions are evaluated for all the seven cases, the
average deviation is -3% with a standard deviation of 5%. These results demonstrate that both methods are
highly accurate, although the number of cases is too few to make any definite conclusions. Hollenbach’s min-
imum power prediction deviates on average -13% for the three applicable cases, and -16% for all the cases
in the study. As expected, the mean prediction by Hollenbach is more accurate, and the results confirm the
decision of not including the minimum estimate in the calculated mean.
Guldhammer-Harvald’s power prediction was only valid for two cases and deviated on average by 31% with a
standard deviation of 5% for the respective cases. However, the updated Guldhammer-Harvald did not apply
100
Chapter 7. Discussion 7.2 Validation Results
to the particular ship types (cargo ships and chemical tankers). Hence the corrections by Kristensen et al.
(2017) are not applied in the calculations. The results indicate that the original Guldhammer-Harvald method
significantly overestimates the power. The updated Guldhammer-Harvald procedure applies to the ship types
in Case 1, 6 and 7, the two container ships and the bulk carrier. For these cases, the average deviation of
the predicted power is -8% with an average standard deviation of 5%. Hence there is a significant effect of
including the modernised corrections in the procedure. Although the results are underestimated, the deviation
is reduced relative to the original method. However, both the container ships and the bulk carrier have ship
dimensions outside the valid range for Guldhammer-Harvald. As a result, the modernised procedure did not
apply to any of the ships in the case study.
There were only three cases where more than one calm water method was applicable, and the mean power
prediction was calculated. On average, for the respective cases, the deviation is 8% with a standard deviation
of 3%. The inclusion of Guldhammer-Harvald contribute to significant deviations in the calculated mean, and
as a result, the individual performance of Hollenbach (mean) and Holtrop-Mennen is better. The results of
the case study suggest that Guldhammer-Harvald is less reliable than Hollenbach and Holtrop-Mennen, even
with the corrections by Kristensen et al. (2017). Based on these results, it should be considered to exclude
the method, or at least only include the updated procedure restricted to tankers, bulk carriers and containers.
This is expected to increase the accuracy of the predictions. Further, excluding Guldhammer-Harvald from
the model will not affect the applicable range for fleet-wide calculations.
Overall, Hollenbach (mean) and Holtrop-Mennen provide accurate results and can apply to a large range
of ships, 72.8% of the current fleet. Hollenbach includes the most restricted applicable range, despite this,
the method provides accurate predictions for the cases outside the valid range. The valid range for ship
dimensions defined by Hollenbach (1997) correspond to the mean value of the ships analysed, ±1.5 stan-
dard deviation. In future work, it may be reasonable to increase the range and allow for larger applicability.
The mean powering prediction will be further improved by excluding Guldhammer-Harvald, although it may
be reasonable to only apply Hollenbach (mean) and use Holtrop-Mennen in cases where Hollenbach is not
applicable.
The performance of the parameter estimates influences the power predictions significantly in the case study.
Among the estimated parameters, the propeller diameter deviates on average -8 % with a standard deviation
on 29% and introduces the most substantial error in the calculations. In cases where the propeller diameter
is significantly underestimated, the propulsive efficiency is underpredicted, and the final power prediction
is correspondingly overestimated. This is a weakness in the model. However, the estimation formulas for
propeller diameter are based on maximum draught, and in the case study, all vessels have design draught or
ballast draught. Fleet-wide calculations apply maximum draught from Sea-web, which is expected to provide
less underpredicted propeller diameter estimates.
7.2.2 In-service Powering Performance

The validation against in-service measurements demonstrates good prediction accuracy for exact loading
conditions, with some variations between the different added resistance methods. The computed RMSE
shows that the applied loading condition affects the power predictions more than varying the added resis-
tance methods. Sea-web provides maximum draught and displacement for all ships, which is not expected to
be a representative loading condition for the fleet at all times. For future work, this should be further investi-
gated and corrected to achieve accurate fleet-wide powering predictions.
For both the cargo ship and the container ship, Townsin & Kwon underestimates the added resistance. The
power predictions are improved in both cases by including the additional methods to predict resistance in wind
and waves. In the model, the mean resistance is computed for the valid methods, and the RMSE for each
combination shows that the prediction accuracy is improved by excluding Townsin & Kwon. A recommendation
for improving the model is to exclude Townsin & Kwon, and further develop a parameter estimate for LBW L in
STA-1 to apply this method. Besides, the parameter estimate for the transverse projected area AV T should
be improved as it shows a significant variability, with an average deviation of -3% and standard deviation of
45%.
101
Chapter 7. Discussion
7.2.3 Comparing the New and Current MariTEAM model

The powering performance of the new and current MariTEAM model is compared, including the parameter
estimates. Table 6.45 summarises the mean deviations for resistance, propulsive efficiency and power in the
validation study. As expected, Hollenbach’s minimum resistance prediction is close to the current MariTEAM
model’s. However, for most cases, the propulsive efficiency is underestimated in the current MariTEAM model,
which result in a less underestimated power. The results indicate that the Wageningen-B series procedure is
more accurate than the propulsive efficiency procedure by Lindstad et al. (2011), if the propeller diameter is
properly estimated. However, the Lindstad formula applies a constant propeller efficiency, which is a better
prediction if the propeller diameter is inaccurately predicted.
Overall, the current MariTEAM model underestimates the calm water power. The mean deviation is, however,
less than for Guldhammer-Harvald. Based on these results, and the results for exact input parameters, it is
recommended to apply the new model, including Hollenbach (mean) and Holtrop-Mennen. A larger validation
study is needed to make any definite conclusions. However, the seven cases investigated represent various
ship types, and there is a clear tendency of more accurate results with Hollenbach’s mean procedure and
Holtrop-Mennen relative to the other methods. In addition, it is recommended to replace the current Townsin
& Kwon added resistance procedure by the four suggested methods, as argued in Section 7.2.2.
102
Chapter 8
Conclusion
A new ship powering performance method for the MariTEAM model is developed and validated in the current
work. The method requires few input parameters and can provide fleet-wide powering predictions in realistic
sea-states. Based on reviewing the current state of knowledge in the field of powering predictions, the most
relevant empirical methods were implemented in the model.
The performance of each method was assessed and validated in a case study of seven different vessels.
Based on the study, it is recommended to apply Holtrop-Mennen and Hollenbach (mean) for calm water pow-
ering predictions. On average, for all the cases, these methods provided powering predictions with a mean
deviation of ± 3 % for exact input parameters, which is a significant improvement to the current MariTEAM
model. Further, by including both methods, 72.8% of the fleet is valid to analyse. Alongside these methods,
Guldhammer-Harvald was suggested, including the updated version of the procedure by Kristensen et al.
(2017). However, the procedure showed a significantly higher deviation, on average, 31 % for the two cases
where it was applicable, and 11 % if all cases are included. Ideally, more cases should be investigated to
make definite conclusions. However, the results demonstrate a clear tendency for seven diverse ship types
which represent large segments of the fleet.
The recommended methods for added resistance in wind and waves include STAJIP wind, Blendermann
(wind) and both STAWAVE procedures. In-service validation of two ships demonstrates an improved powering
prediction for these methods relative to the Townsin & Kwon method currently applied in the current MariTEAM
model. Therefore, it is recommended to exclude Townsin & Kwon. The prediction of the propulsive efficiency
is further improved for the current model by implementing the Wageningen B-series procedure. On average
for all the cases, the procedure provided propulsive efficiency predictions with a mean deviation of 3 %, when
Hollenbach and Holtrop-Mennen were applied. The new model is subdivided into modules, which makes it
easy to apply the current recommendations and implement new procedures in future work.
103
Chapter 8. Conclusion
104
Chapter 9
Recommendations for Further Work
In addition to the specific recommendations presented in the conclusion, the following suggestions are given
for further work.
Parameter estimates
As the case study demonstrates, the accuracy of the parameter estimates affects the performance of the
powering prediction. A new parameter estimate for LBW L (the length of the bow to 95 % of the maximum
breadth on the waterline), is required for STA-1. This may be developed by investigating typical bow designs
for the main ship types in the fleet. Further, the estimate of the transverse projected area AV T deviates
significantly and should be improved. In addition, the propulsive efficiency computed by the Wageningen
B-series is sensitive to errors in the estimated propeller diameter, which should be further investigated. As
discussed, the estimate is expected to improve for maximum draught input, although this must be confirmed.
Loading conditions
Both the calm water validation study and the in-service study demonstrated significant variations in the pow-
ering predictions for varying loading conditions. Applying a maximum loading condition to all vessels in the
fleet is expected to overpredict fleet-wide powering and fuel consumption significantly. As Sea-web provides
maximum draught and displacement, a fleet-wide study of typical loading conditions is necessary to develop a
correction to the current powering predictions. AIS data containing draught may be utilised in the investigation.
Valid range for Hollenbach
The valid range for Hollenbach (1997) is defined in terms of several dimensionless ship characteristics. The
range for each dimension is defined as the mean value of all the ships analysed in Hollenbach’s study, ±
1.5 standard deviation. The range is significantly more restrictive than the range defined in Holtrop-Mennen.
Results from the case study indicate that Hollenbach provides accurate powering predictions for vessels
outside the range. Although the case study is limited to seven ships, it may be reasonable to increase the
range and allow for larger applicability in future work.
105
106
Bibliography
Alte, R. and Baur, M. V. (1986). Propulsion. Handbuch der Werften. Hansa, 18(S 132).
Andersen, P. and Guldhammer, H. E. (1986). A Computer-Oriented Power Prediction Procedure. In Inter-

national Conference on Computer Aided Design, Manufacture and Operation in the Marine and Offshore
IndustriesComputational Mechanics Publications.
Auf’m Keller, W. (1973). Extended diagrams for determining the resistance and required power for single-
screw ships. International Shipbuilding Progress, 20:133–142.
Bertram, V. (2012). Practical Ship Hydrodynamics. Elsevier, 2 edition.
Bertram, V. and Wobig, M. (1999). Simple empirical formulae to estimate main form parameter. Schiff Hafen,
11(1):118–121.
Birk, L. (2019). Fundamentals of Ship Hydrodynamics. Wiley, 1 edition. Publication Title: ISBN: 978-1-118-
85551-5.
Blendermann, W. (1986). The wind forces on ships. In Rep. 467. Institut für Schiffbau Hamburg.
Blendermann, W. (1994). Parameter identification of wind loads on ships. Journal of Wind Engineering and
Industrial Aerodynamics, 51(3):339–351. ISBN: 0167-6105 Publisher: Elsevier.
Blendermann, W. (1995). Estimation of wind loads on ships in wind with a strong gradient. Proceedings of the
14th International Conference on Offshore Mechanics and Arctic Engineering (OMAE), New York, ASMW,
1-A:271–277.
Blendermann, W. (2004). Ships may encounter high wind loads—a statistical assessment. Proceedings
of the Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime Environment,
218(1):1–10. Publisher: SAGE Publications.
Boswell, R. J. (1971). Design, cavitation performance, and open-water performance of a series of research
skewed propellers. Technical report, David W Taylor Naval Ship Research and Development Center
Bethesda MD.
Bouman, E. A., Lindstad, H. E., and Stromman, A. H. (2016). Life-Cycle Approaches for Bottom-Up As-
sessment of Environmental Impacts of Shipping. In SNAME Maritime Convention. The Society of Naval
Architects and Marine Engineers.
Breslin, J. P. and Andersen, P. (1994). Hydrodynamics of Ship Propellers. ISBN: 9780521413602

9780521574709 9780511624254 Library Catalog: www.cambridge.org Publisher: Cambridge University
Press.
Brix, J., Blendermann, W., Grensemann, K., Norrbin, N. H., Pieper, W., Schwanecke, H., Sharma, S. D.,
Söding, H., Wagner, B., and Weiss, F. (1993). Manoeuvring technical manual, volume 36. Seehafen
Verlag, Hamburg. ISBN: 0036-603X.
Clarksons (2020). Sea/net - Clarksons Research Portal.
Co, W. a. (1985). Prediction of Windloads on Large Liquified Gas Carriers by Witherby & Co. London for
Oil Companies International Marine Forum (OCIMF) and Society of International Gas Tanker & Terminal
Operators Ltd. (SIGTO). Technical report, Witherby & Co.
107
Dale, T. (2020). Development of Simplified Methods for Ship Powering Performance Calculations. Project
Thesis, NTNU, Trondheim.
Danckwardt, E. C. M. (1969). Ermittlung des Widerstandes von Frachtschiffen und Hecktrawlern beim Entwurf.
Schiffbauforschung.
Digernes, T. (1982). An analytical approach to evaluating fishing vessel design and operation. Norges tekniske
høgskole.
ECMWF (2014). ERA-Interim. Library Catalog: www.ecmwf.int.
Edenhofer, O. (2014). Climate change 2014: mitigation of climate change: Working Group III contribution
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Technical report,
Cambridge University Press, New York, NY. OCLC: ocn892580682.
EU-council (2019). CO2 emissions from ships: Council agrees its position on a revision of EU rules. Library
Catalog: www.consilium.europa.eu.
Faltinsen, O. (1993). Sea loads on ships and offshore structures, volume 1. Cambridge university press.
Faltinsen, O. M., Minsaas, K., Skjørdal, O, S., and Liapis, N. (1980). Prediction of Resistance and Propulsion of
a Ship in a Seaway. In Proceeding of 13th Symposium on Naval Hydrodynamics. In Prediction of Resistance
and Propulsion of a Ship in a Seaway. In Proceeding of 13th Symposium on Naval Hydrodynamics, pages
505–529.
Fujiwara, T. (2005). A New Estimation Method of Wind Forces and Moments Acting on Ships on the Basis
of Physical Components Models. Journal of the Japan society of naval architects and ocean engineers,
2:243–255.
Fujiwara, T., Kitamura, F., Ueno, M., and Sogihara, N. (2017). Estimation of above water structural parameters
and wind loads on ships. Ships and Offshore Structures, 12(8):1100–1108.
Gawn, R. and Burrill, L. (1957). Effect of Cavitation on the Performance of a Series of 16 in. Model Propellers.
In Meeting of the Institution of Naval Architects, TINA, London, 1954, Published in: RINA Transactions:
1957-32, Paper 3.
Gawn, R. W. L. (1953). Effect of pitch and blade width on propeller performance. In Autumn Meeting of the
Institution of Naval Architects, TINA, Rome, Italy, 1953, pp. 157-193, RINA Transactions 1953-09.
Gertler, M. (1954). A reanalysis of the original test data for the Taylor Standard Series. Technical report,
David Taylor Model Basin Washington.
Guldhammer, H. E. and Harvald, S. A. (1974). Ship Resistance - Effect of form and principal dimensions.
(Revised). Danish Technical Press, Danmark, Danmarks Tekniske Hojskole, kademisk Forlag, St. kan-
nikestrade 8, DK 1169 Copenhagen.
Gupta, P., Steen, S., and Rasheed, A. (2019). Big Data Analytics As a Tool to Monitor Hydrodynamic Per-
formance of a Ship. In Volume 7A: Ocean Engineering, page V07AT06A059, Glasgow, Scotland, UK.
American Society of Mechanical Engineers.
Harvald, S. A. (1992). Resistance and propulsion of ships. Krieger Publishing Company.
Helm, G. (1964). Systematic (model) investigations on the resistance of boats and small ships. Hansa,
101(22):p. 2179.
Hoegh-Guldberg, O., Lovelock, C., Caldeira, K., Howard, J., Chopin, T., and Gaines, S. (2019). The Ocean as
a Solution to Climate Change: Five Opportunities for Action. Technical report, World Resources Institute,
Washington DC. Publisher: World Resources Institute.
Hollenbach, K. U. (1997). Beitrag zur abschätzung von widerstand und propulsion von ein-und zweis-
chraubenschiffen im vorentwurf. PhD thesis, Inst. für Schiffbau.
Hollenbach, K. U. (1998). Estimating resistance and propulsion for single-screw and twin-screw ships. Schiff-
stechnik, 45(2):72.
108
Holtrop, J. (1977). A Statistical Analysis of Performance Test Results. International Shipbuilding Progress,
24(270).
Holtrop, J. (1984). A statistical re-analysis of resistance and propulsion data. International shipbuilding
progress, 31(363):272–276. ISBN: 0020-868X.
Holtrop, J. and Mennen, G. (1982). An approximate power prediction method. International Shipbuilding
Progress, 29(335):166–170.
IHS (2020). Maritime Sea-web Online Ship Register | IHS Markit. Library Catalog: ihsmarkit.com.
IMO (2020). Automatic Identification Systems (AIS).
IndEcol (2019). IndEcol - Industrial Ecology Programme - NTNU.
ISO Technical Committee (2015). ISO 15016:2015 Ships and marine technology — Guidelines for the as-
sessment of speed and power performance by analysis of speed trial data.
ITTC (2014). ITTC Analysis of Speed/Power Trial Data ( ITTC - Recommended Procedures 7.5-04-01-01.2).
ITTC (2017). 1978 ITTC Performance Prediction Method (ITTC - Recommended Procedures 7.5-02-03-01.4).
ITTC (2017). Description and Rules of the ITTC.
ITTC (2018). ITTC Calculation of the weather factor (fw) for decrease of ship speed in wind and waves ( ITTC
- Recommended Procedures 7.5-02-07-02.8).
ITU (2018). List of Ship Stations and Maritime Mobile Service Identity Assignments (List V) - Compilation of
amendments.
Jalkanen, J.-P., Johansson, L., Kukkonen, J., Brink, A., Kalli, J., Stipa, T., and Kerminen, V.-M. (2012). Exten-
sion of an assessment model of ship traffic exhaust emissions for particulate matter and carbon monoxide.
Atmospheric Chemistry & Physics, 12(5). ISBN: 1680-7316.
Johansson, L., Jalkanen, J.-P., and Kukkonen, J. (2017). Global assessment of shipping emissions in 2015
on a high spatial and temporal resolution. Atmospheric Environment, 167:403–415.
Kim, Y. (2020). PhD Project (in-progress): Develop new models for hull resistance, engine emissions, drive-
line, propeller efficiency and fuel production that enables efficient fleetwide modelling (CLIMMS WP2).
Kristensen, H. O., Lützen, M., and Bingham, H. (2017). Prediction of Resistance and Populsion Power of
Ships (Work Package 2, Report no. 4). Technical report, Technical University of Denmark.
Kwon, Y. J. (2008). Speed loss due to added resistance in wind and waves. The Naval Architect, RINA,
3:14–16.
Lackenby, H. (1963). The effect of shallow water on ship speed. Shipbuilder and Marine Engineer, 70:446–
450.
Lap, A. J. W. (1954). Diagrams for determining the resistance of single-screw ships. International Shipbuilding
Progress, 1(4):179–193. ISBN: 0020-868X Publisher: IOS Press.
Lindstad, H., Asbjørnslett, B. E., and Jullumstrø, E. (2013). Assessment of profit, cost and emissions by
varying speed as a function of sea conditions and freight market. Transportation Research Part D: Transport
and Environment, 19:5–12. ISBN: 1361-9209 Publisher: Elsevier.
Lindstad, H., Asbjørnslett, B. E., and Strømman, A. H. (2011). Reductions in greenhouse gas emissions and
cost by shipping at lower speeds. Energy Policy, 39(6):3456–3464. Publisher: Elsevier.
Lindstad, H., Sandaas, I., and Steen, S. (2014). Assessment of profit, cost, and emissions for slender bulk
vessel designs. Transportation Research Part D: Transport and Environment, 29:32–39. ISBN: 1361-9209
Publisher: Elsevier.
Liu, S. and Papanikolaou, A. (2016). Fast approach to the estimation of the added resistance of ships in head
waves. Ocean Engineering, 112:211–225.
109
Lloyd’s (2020). Shipping and maritime intelligence - Lloyd’s List Intelligence.
Lu, R., Turan, O., Boulougouris, E., Banks, C., and Incecik, A. (2015). A semi-empirical ship operational per-
formance prediction model for voyage optimization towards energy efficient shipping. Ocean Engineering,
110:18–28.
Lützen, M. and Kristensen, H. O. H. (2012). A Model for Prediction of Propulsion Power and Emissions –
Tankers and Bulk Carriers. In World Maritime Technology Conference WMTC 2012, page 13.
MAN (2018). Marine Engines & Systems.
MARIN (2006). Sea Trial Analysis (STA) - JIP. Issued Reports.
MarineTraffic (2018). Four ways MarineTraffic ensures AIS data accuracy. Library Catalog:
www.marinetraffic.com Section: AIS – essential knowledge.
MARINTEK (2020). MARINTEK, Norwegian Marine Technology Research Institute, Trondheim, Norway.
Molland, A. F. (2011). The maritime engineering reference book: a guide to ship design, construction and
operation. Elsevier.
Molland, A. F., Turnock, S. R., and Hudson, D. A. (2017). Ship resistance and propulsion. Cambridge university
press.
Moor, D. I., Parker, M. N., and Pattullo, R. N. M. (1961). The BSRA Methodical Series–An Overall Presentation.
Transactions of the Royal Institution of Naval Architects, 103(329-419):1. Publisher: The Institution.
Muri, H., Strømman, A. H., Ringvold, A. L., Lonka, R., and Bouman, E. (2019a). Influence of weather on
emissions from the global shipping fleet. In Geophysical Research Abstracts, volume 21.
Muri, H., Strømman, A. H., Ringvold, A. L., Lonka, R., Lindstad, E., and Bouman, E. A. (2019b). A new
emission inventory of the global maritime fleet; the effect of weather. AGUFM, 2019:A21W–2637.
Newton, R. and Rader, H. (1961). Performance data of propellers for high speed craft. Admiralty Experiment
Works, Haslar, UK, Published by: The Royal Institution of Naval Architects, RINA Transactions 1961-07,
Volume 103, No. 2, Quarterly Transactions, pp. 93-129.
Olmer, N., Comer, B., Roy, B., Mao, X., and Rutherford, D. (2017). Greenhouse gas emissions from global
shipping, 2013-2015 Detailed methodology. ICCT (The International Council on Clean Transportation),
page 52.
Oosterveld, M. W. C. (1970). Wake adapted ducted propellers. Technical report, NSBM.
Oosterveld, M. W. C. and van Oossanen, P. (1975). Further computer-analyzed data of the Wageningen
B-screw series. International shipbuilding progress, 22(251):251–262. ISBN: 0020-868X Publisher: IOS
Press.
Papanikolaou, A. (2014). Ship Design: Methodologies of Preliminary Design. Springer. Google-Books-ID:

WACKBAAAQBAJ.
Ringvold, A. (2017). Prospective Life Cycle Assessment of Container Shipping. PhD thesis, NTNU.
Schneekluth, H. and Bertram, V. (1998). Ship design for efficiency and economy. Butterworth-Heinemann,
Oxford, 2nd ed. edition. Pages: 220.
SFI, S. M. (2016). Sub Project 4 - Performance in a Seaway. Library Catalog: www.smartmaritime.no.
Smith, Jalkanen, J. P., Anderson, B. A., Corbett, J. J, Faber, J., Hanayama, S., O’Keeffe, E., Parker, S.,
Johansson, L., Aldous, L., Raucci, C., and Traut, M. (2014). Third IMO Greenhouse Gas Study 2014.
Technical report, International Maritime Organization, London.
Steen, S. (2011). TMR4247 Marin teknikk 3-Hydrodynamikk: Motstand og propulsjon, propell og foilteori.
volume UK-2011-99. Department of marine technology, Trondheim.
110
Steen, S. (2020). Personal communication with Sverre Steen, Professor in marine hydrodynamics, Norwegian
University of Science and Technology.
Steen, S., Koushan, K., and Minsaas, K. (2016). Ship Resistance - Lecture notes TMR4220 Naval hydrody-
namics. Department of Marine Technology, NTNU.
Tillig, F. (2017). A Generic Model for Simulation of the Energy Performance of Ships – from Early Design to
Operational Conditions. PhD thesis, Chalmers University of Technology.
Tillig, F. (2020). Simulation model of a ship’s energy performance and transportation costs. PhD thesis,
Chalmers University of Technology.
Tillig, F., Ringsberg, J. W., Mao, W., and Ramne, B. (2018). Analysis of uncertainties in the prediction of ships’
fuel consumption – from early design to operation conditions. Ships and Offshore Structures, 13(sup1):13–
24. Publisher: Taylor & Francis _eprint: https://doi.org/10.1080/17445302.2018.1425519.
Todd, F. H. (1957). Series 60-The Effect upon Resistance and Power of Variation in Ship Propotions. Trans.
SNAME, 65:445–589.
Townsin, R. L. and Byrne, D. (1980). Speed, power and roughness: the economics of outer bottom mainte-
nance. Technical report, Royal Institution of Naval Architects. ISBN: 0306-0209.
Townsin, R. L. and Kwon, Y. J. (1983). Approximate formulae for the speed loss due to added resistance in
wind and waves. Transactions of the Royal Institution of Naval Architects.
Townsin, R. L. and Kwon, Y. J. (1993). Estimating the influence of weather on ship performance. Transactions
of the Royal Institution of Naval Architects.
Townsin, R. L. and Mosaad, M. A. (1985). The ITTC Line - Its genesis and Correlation Allowance. Naval
Arcitech.
Troost, L. (1951). Open water test series with modern propeller forms. Netherlands Ship Model Basin, Wa-
geningen, Paper presented at: The North East Coast Institution of Engineers and Shipbuilders in Newcastle
upon Tyne, UK, Institution Transactions Volume 67.
UNCTAD (2018). Review of Maritime Transport - United Nations Conference on Trade and Development,
UNCTAD/RMT/2018. In Review of Maritime Transport - United Nations Conference on Trade and Develop-
ment, UNCTAD/RMT/2018, New York and Geneva. United Nations Publications. OCLC: 1083029692.
Valanto, P. and Hong, Y. (2017). Wave Added Resistance and Propulsive Performance of a Cruise Ship in
Waves. In Proceedings of the Twenty-seventh (2017) International Ocean and Polar Engineering Confer-
ence, page 9. ISOPE.
Van Oortmerssen, G. (1971). A power prediction method and its application to small ships. International
Shipbuilding Progress, 18(207):397–415. ISBN: 0020-868X Publisher: IOS Press.
Wartsila (2019). Ship Resistance.
Watson, D. G. (1998). Practical ship design, volume 1. Elsevier.
111
112
Appendix A
Hollenbach Coefficients
CRstd = b11 + b12 FN + b13 FN2

+ b21 + b22 FN + b23 FN2 CB (A.1)
+ b31 + b32 FN + b33 FN2 2
CB
Figure A.1: Coefficients for computing CR in Hollenbach as presented by Birk (2019)
I
II
Appendix B
Guldhammer-Harvald Coefficients
Kristensen et al. (2017) presents coefficients for the original CR curves in Guldhammer-Harvald, based on
extensive regression analysis. The details are outlined in the following. Note that in the current work, the
residuary resistance coefficient without corrections is denoted CR,diagram .
103 · CR,diagram = E + G + H + K
where:
⇣ ⌘
E = A0 + 1.5 · Fn1.8 + A1 · FnN1 · 0.98 + 2.5
(M 2)4 + (M 5)4 · (Fn 0.1)4
A0 = 1.35 0.23 · M + 0.012 · M2
A1 = 0.0011 · M9.1
N1 = 2 · M 3.7
(B.1)
B1 ·B2
G= B3
B1 = 7 0.09 · M2
2
B2 = (5 · CP 2.5)
1.5
B3 = 600 · (Fn 0.315)2 + 1
H = exp (80 · (Fn (0.04 + 0.59 · CP ) 0.015 · (M 5)))
K = 180 · Fn3.7 · exp (20 · CP 16)
Hull-propeller interaction parameters
Harvald (1992) present estimates for t and w as outlined by Lützen and Kristensen (2012), presented in
Equation B.2
⇣ ⌘
B Dp
w = w1 L , CB + w2 (form, CB ) + w3 L
⇣ ⌘ (B.2)
B D
t = t1 L , CB + t2 (form) + t3 Lp
The parameters of Equation B.2 can be approximated in accordance with diagrams from Harvald (1992).
Lützen and Kristensen (2012) presents approximated values from these diagrams by the following regression
formulas. Wake fraction is estimated by Equation B.3
w = w1 + w2 + w3 (B.3)
III
where
b
w1 = a + c·(0.98 CB )3 +1
0.025·Fa
w2 = 100·(CB 0.7)2 +1
0.00756
w3 = 0.18 + 0.00756
+0.002
and w3  0.1
L
and
0.1·B
a= L + 0.149
0.05·B
b= L + 0.449
5027·B B 2
c = 585 L + 11700 · L
Fa is a form parameter defined as [-2,0,2] respectively for the hull shapes [U, N, V]. The thrust deduction is
estimated by Equation B.4
t = t 1 + t2 + t 3 (B.4)
where
e
t1 = d + f ·(0.98 CB )3 +1
t2 = 0.01 · Fa
⇣ ⌘
Dp
t3 = 2 · L 0.04
and
0.625·B
d= L + 0.08
0.25·B
e = 0.165 L
8060·B B 2
f = 525 L + 20300 · L
IV
Appendix C
Holtrop-Mennen Coefficients
Full scale wake fraction coefficients
Coefficients for the full scale wake fraction in Equation 3.22, as presented by Birk (2019).
8 ⇣ ⌘
S B
>
> if B/TA  5
< LW L D TA
c8 = ⇣ ⌘ (C.1)
> S 7 TB 25
>
: A
⇣ ⌘ if B/TA > 5
LW L D TB 3
A
8
< c8 if c8  28
c9 = (C.2)
: 32 16
if c8 > 28
c8 24
8 TA
< D if TA /D  2
c11 = (C.3)
: TA 3
0.0833333 D + 1.33333 if TA /D > 2
8 0.12997 0.11056
< (0.95 CB ) (0.95 CP ) if CP  0.7
c19 = (C.4)
: 0.18567
0.71276 + 0.38648CP if CP > 0.7
(1.3571 CM )
c20 = 1 + 0.015Cstern
(C.5)
CP 1 = 1.45CP 0.315 0.0225`CB
Holtrop’84 hull form factor

"✓ ◆1.06806 ✓ ◆0.46106 ✓ ◆0.121563 ✓ ◆0.36486 #
B T LW L L3W L 0.604247
k= 0.07+0.487118c14 (1 CP )
LW L LW L LR V
(C.6)
Wave resistance coefficients
Coefficients for computing the wave resistance if F n  0.4 according to Birk (2019).
V
8 ⇣ ⌘(1/3)
>
> B
0.229577 LW if B/LW L  0.11
>
>
< L
c7 = B
if 0.11 < B/LW L  0.25 (C.7)
>
> LW L
>
>
:
0.5 0.0625 LW
B
L
if B/LW L > 0.25
✓ ◆1.07961
T 1.37565
c1 = 2223105c3.78613
7 (90 iE )
B
A1.5
BT
c3 = 0.56 ⇥ p ⇤
BT 0.31 ABT + TF hB (C.8)
p
c 2 = e( 1.89 c3 )
AT
c5 = 1 0.8
BT CM
8 L3W L
>
> 1.69385 if  512
>
> V
>
< LW L
c15 = 1.69385 + V (1/3) 8
if 512 <
L3W L
 1726.91 (C.9)
>
> 2.36 V
>
>
>
: L3W L
0 if V > 1726.91
8
< 8.07981CP 13.8673CP2 + 6.984388CP3 if CP  0.8
c16 =
: 1.73014 0.7067CP if CP > 0.8 (C.10)
d= 0.9
8
< 1.446CP 0.03 LW
B
L
if LW L
B  12
= (C.11)
: 1.446C 0.36 if LW L
> 12
P B
(1/3)
m1 = 0.0140407 LW
T
L
1.75254 VLW L B
4.79323 LW L
c16
(C.12)
3.29
m4 = 0.4c15 e( 0.034Fn )
VI
Appendix D
Sea-web Parameters
VII
Aux. Engine builder Engine design IMO chemical class II
Aux. engine design Engine model* IMO chemical class III
Aux. engine model Engine stroke Inert gas system
Aux. engine stroke type* Engine stroke type* Insulated capacity
Aux. Engine Total kW* Engine type Keel Laid
Aux engines number Engines number Keel to Mast height
Bale Engines RPM* Last update
Bollard pull Est. Comp. Date Launch date
Bow discharge Ex-flag LDT*
Bow loading ExName Leadship
Bow stoppers Fairplay ID Length
Bow stoppers no Fishing No. Length BP*
Bow stoppers swl Flag Length Registered
Bow to centre manifold Flash Point < 60 Lines per side
Breadth* Flash point > 60 Liquid
Breadth extreme Formula DWT Manifold
Breadth moulded Fuel capacity 1 Maniful diameter
Built* Fuel capacity 2 Marpol 13G Phaseout
Cabins Fuel consumption Main Mid point manifold aft
Callsign Engines ballast
Cancel data Fuel Consumption Total Mid point manifold aft
Cargo tank coating Fuel Type 1* laden
Cargo tanks Fuel Type 2 Mid point manifold aft
Cars Gas Capacity light
Class Gear No Largest Mid point manifold fwd
Clean ballast Gear SWL Largest ballast
Closed loading Gir Type Largest Mid point manifold fwd
Cont hull survey Gearless laden
Converted (last/only) Generators kW Mid point manifold fwd
Country of build Grades light
Crew Grain MMSI*
Crude oil washing Group Owner Movements
Deadweight* Group Owner Code Newbuilding price
Decks Group Owner Control NRT
Delivery date Group Owner Domicile Official No.
Depth Group Owner Operator
Displacements Registration Operator Code
DOC company GT* Operator Control
DOC company code Hatches Operator Domicile
DOC control Heating Coil Material Operatior Registration
DOC domicile Heating Coils Order Date
DOC registration Heating Coils Slop P and I Club
Docking survey Holds Parallel Body Length
Draught* Hull material Ballast
Engine bore Hull Type Parallel Body Length
Engine builder Ice Capable Laden
Engine cylinders* IMO chemical class I Parallel Body Length Light
VIII
Passengers Shipmanager domicile
PCNT Shipmanager registration *In the current
Permanent Ballast Slop capacity MariTEAM database
Photo Slop tank coating
Port Slop tanks
Propeller Manufacturer SMC Auditor
Propeller Type SMC Date Expires
Propulsion Type* SMC Date Issued
Propulsion Units SMC Issuer
Recycling Arrival Special Survey
Recycling Commenced SPM Equipped
Recycling Country Standard Design
Recycling LDT Price Statcode 5
Recycling Price
Recycling Yard Status
Reefer Points* Status Date
Registered Owner Stern Discharge
Registered Owner Code Stern Loading
Registered Owner Sub Contractor
Control Sub Contractor Yard No
Registered Owner Tank Heat Exchanger
Domicile Material
Registered Owner Tank Heat Exchangers
Registration Tanks
Retirement Technical Manager
RORO Lanes Height Technical Manager Code
RORO Lanes Length Technical Mgr Control
RORO Lanes Number Technical Mgr Domicile
RORO Lanes Width Technical Mgr
RORO Ramp Position Registration
RORO Ramp SWL Temperature Max
Sale Date Temperature Min
Sale Price (US$) TEU*
SCNT TEU 14
Segregated ballast Thruster Largest kW
Segregated ballast – Thruster Largest Type
Protected Thrusters Number
Segregated ballast Thrusters Total kW
Capacity Total HP Main Eng
Service Speed* Total kW Main Eng*
Ship Type* TPC
Ship Type Group* TVE Expiry Date
Shipbuilder Vapour Recovery
Shipbuilder code Yard No.
Shipmanager Year
Shipmanager code
Shipmanager control
IX
X
Appendix E
Blendermann Wind Coefficients
Wind forces and moments in dimensionless form are presented in the diagrams by Blendermann (Brix et al.,
1993). The longitudinal dimensionless wind resistance coefficient CX is retrieved from the diagram and
tabulated for each ship. Note that the diagrams present the wind resistance coefficient as CX according to
the definitions in the current work.
E.1 Car Carrier
Table E.1: Car carrier wind coefficients
Angle of Attack CX
0 -0.55
10 -0.65
20 -0.65
30 -0.60
40 -0.40
50 -0.10
60 0.18
70 0.15
80 0.00
90 -0.20
100 -0.30
110 -0.25
120 -0.12
130 0.20
140 0.50
150 0.70
160 0.80
170 0.74
180 0.60
XI
E.2 Container Vessel
Table E.2: Container vessel wind coefficients
Angle of Attack CX
0 -0.55
10 -0.47
20 -0.46
30 -0.52
40 -0.52
50 -0.44
60 -0.34
70 -0.26
80 -0.11
90 -0.03
100 0.08
110 0.22
120 0.44
130 0.58
140 0.60
150 0.63
160 0.58
170 0.54
180 0.40
E.3 Tanker (loaded)
Table E.3: Tanker (loaded) wind coefficients
Angle of Attack CX
0 -0.90
10 -0.87
20 -0.95
30 -0.95
40 -0.85
50 -0.72
60 -0.51
70 -0.29
80 -0.03
90 0.10
100 0.26
110 0.32
120 0.42
130 0.51
140 0.57
150 0.62
160 0.61
170 0.60
180 0.55
XII
E.4 Cargo Vessel
Table E.4: Cargo vessel wind coefficients
Angle of Attack CX
0 -0.65
10 -0.67
20 -0.77
30 -0.77
40 -0.70
50 -0.52
60 -0.37
70 -0.21
80 -0.05
90 0.15
100 0.22
110 0.29
120 0.29
130 0.37
140 0.47
150 0.57
160 0.65
170 0.59
180 0.60
E.5 Passenger Ship
Table E.5: Passenger ship wind coefficients
Angle of Attack CX
0 -0.40
10 -0.37
20 -0.35
30 -0.27
40 -0.08
50 0.09
60 0.04
70 0.00
80 -0.01
90 -0.03
100 -0.08
110 -0.08
120 -0.16
130 -0.18
140 -0.06
150 0.21
160 0.33
170 0.38
180 0.37
XIII
E.6 Offshore Supply Vessel
Table E.6: Offshore supply vessel wind coefficients
Angle of Attack CX
0 -0.55
10 -0.55
20 -0.51
30 -0.50
40 -0.43
50 -0.33
60 -0.25
70 -0.17
80 -0.12
90 -0.10
100 -0.07
110 0.07
120 0.37
130 0.59
140 0.71
150 0.79
160 0.75
170 0.81
180 0.81
E.7 Ro-Ro/Lo-Lo
Table E.7: Ro-Ro/Lo-Lo wind coefficients
Angle of Attack CX
0 -0.66
10 -0.66
20 -0.74
30 -0.71
40 -0.67
50 -0.55
60 -0.40
70 -0.25
80 -0.14
90 -0.09
100 -0.15
110 -0.05
120 0.00
130 0.26
140 0.52
150 0.81
160 0.88
170 0.79
180 0.70
XIV
E.8 Deep Sea Drilling Vessel
Table E.8: Deep sea drilling vessel wind coefficients
Angle of Attack CX
0 -1.05
10 -1.22
20 -1.39
30 -1.38
40 -1.27
50 -1.08
60 -0.78
70 -0.56
80 -0.34
90 -0.15
100 -0.04
110 0.23
120 0.67
130 1.01
140 1.23
150 1.37
160 1.34
170 1.25
180 1.25
E.9 Research Vessel
Table E.9: Research vessel wind coefficients
Angle of Attack CX
0 -0.55
10 -0.59
20 -0.64
30 -0.59
40 -0.54
50 -0.47
60 -0.40
70 -0.29
80 -0.21
90 -0.18
100 -0.15
110 -0.06
120 0.10
130 0.39
140 0.62
150 0.82
160 0.83
170 0.71
180 0.60
XV
XVI
Appendix F
The New Program (MATLAB code)
Main
%------------------------------------------------------------------------%
% main.m - Empirical ship powering prediction method
%
% Other m-files required: import_ship_data.m, import_weather_data.m
%
% Subfunctions: air_resistance.m, beaufort.m, blendermann_wind.m,
% estimate_hull_char.m, estimate_prop_char.m, GH.m, HM.m, hollenbach.m,
% kwon.m, requirements.m, STA1.m, STA2.m, STAJIP_wind.m, wageningen.m
%
% Inputs (from Sea-web):
% B - Breadth (moulded)
% T - Draught
% dwt - Dead weight
% ldt - Lightship weight
% Loa - Length over all
% Lpp - Length between perpendiculars
% ship_type - Ship type (string)
% D - Depth from keel to uppermost continuous deck
% no_decks - Number of decks
% Pin - Installed power
% ME_n - Main engine rpm
% Vd - Design speed
% bulb - Either 'yes' or 'no'
% teu - TEU if containership
%
% Inputs (Weather):
% Hs - Significant wave height
% Twave - Wave period
% wave_angle - Relative wave direction
% U - Relative wind speed
% U_angle - Relative wind direction
% U_true - True wind speed
%
% Outputs:
% PB - Total brake power
%
% Author: Tone Dale
% Department of Marine technology, NTNU
% email address: tonedale@stud.ntnu.no
% April 2020; Last revision: 14-July-2020
%-------------------------------------------------------------------------%
%----------------------------- Import data -------------------------------%
% Constants
rho = 1026.0210;
rho_air = 1.250; % Dry air 15 degrees
XVII
g = 9.81;
nu = 1.1892E-6;
p_v = 1.6709E3;
% Import ship data from Sea-web

[B,T,D,no_decks,dwt,ldt,Loa,Lpp,Pin,ME_n,Vd,ship_type,sub_type,teu,bulb]...
= import_ship_data();
% Import weather data (and in-service data for validation)

[Vsog,V,Hs,H,wave_angle,Twave,U,U_angle,U_true,P_aux_is,P_shaft_is,...
shaft_rpm_is,ME_load_is,P_DG1_is,P_DG2_is,P_DG3_is]...
= import_weather_data();
%----------------------- MODULE 1: Estimate input ------------------------%
% Estimate missing hull input parameters

[Lwl,Los,Lfn,TA,TF,voldisp,S,SB,Sapp,AVT,AVL,CB,CM,CP,CWP] ...
= estimate_hull_char(B,T,D,no_decks,teu,dwt,ldt,Lpp,ship_type,rho);
% Estimate missing propulsion input parameters

[Dp] = estimate_prop_char(T,ship_type);
%-------------------- MODULE 2: Calm water resistance --------------------%
% Calculate calm water resistance as mean of applicable methods
% Check which methods fulfill the respective requirements

[GH_req,HM_req,HB_req] = requirements(Lpp,Lwl,Los,B,T,TA,CB,CP,Dp,...
voldisp,bulb);
% Guldhammer Harvald (updated by Kristensen 2017)

[RT_GH,etaH_GH,t_GH,w_GH] = GH(V,Vd,voldisp,Lwl,Lpp,Los,B,T,S,AVT,...
CB,CP,Dp,ship_type,sub_type,teu,bulb,rho,g,nu);
RT_GH = RT_GH*GH_req;
etaH_GH = etaH_GH*GH_req;
t_GH = t_GH*GH_req;
w_GH = w_GH*GH_req;
% Holtrop-Mennen
[RT_HM,etaH_HM,t_HM,w_HM] = HM(V,Vd,voldisp,Lwl,Lpp,B,TA,TF,S,Sapp,...
CB,CP,CM,CWP,Dp,rho,g,nu,ship_type,sub_type,AVT,teu);
RT_HM = RT_HM*HM_req;
etaH_HM = etaH_HM*HM_req;
t_HM = t_HM*HM_req;
w_HM = w_HM*HM_req;
% Hollenbach (mean)
[RT_HB,etaH_HB,t_HB,w_HB] = hollenbach(V,Lwl,Lpp,Los,Lfn,B,TA,TF,S,...
CB,Dp,rho,g,nu,ship_type,sub_type,AVT,teu);
RT_HB = RT_HB*HB_req;
etaH_HB = etaH_HB*HB_req;
t_HB = t_HB*HB_req;
w_HB = w_HB*HB_req;
RT = mean_value(RT_HB,RT_GH,RT_HM);
etaH = mean_value(etaH_HB,etaH_GH,etaH_HM);
t = mean_value(t_HB, t_GH, t_HM);
w = mean_value(w_HB, w_GH, w_HM);
XVIII
% Plot results from calm water resistance predictions
plot_calm_water_resistance(RT_HB,RT_GH,RT_HM,RT,V,g,Lpp);
%------------------ MODULE 3: Added weather resistance -------------------%
% Calculate added resistance due to weather
% Wave resistance
% STA 1
R_wave_STA1 = STA1(rho,g,Hs,wave_angle,B,Lpp);
% STA 2
[R_wave_STA2,STA2_req] = STA2(rho,g,Hs,H,B,Lwl,Lpp,Twave,V,CB,T,...
wave_angle);
R_wave_STA2 = R_wave_STA2.*STA2_req;
R_wave = mean_value(R_wave_STA1,R_wave_STA2,0);
% Wind resistance
% STA-JIP wind (ISO 15016)

R_wind_STAJIP = STAJIP_wind(ship_type,sub_type,AVT,rho_air,U,...
U_angle,Vsog);
% Blendermann wind
R_wind_blendermann = blendermann_wind(ship_type,sub_type,AVT,...
rho_air,U,U_angle,V);
R_wind = mean_value(R_wind_STAJIP,R_wind_blendermann,0);
% Total added resistance
% Kwon
R_added_kwon = kwon(CB,Lwl,V,H,wave_angle,U_true,U_angle,g,voldisp,...
ship_type,RT);
% Find mean added resistance

R_added = mean_value(R_wave+R_wind,R_added_kwon,0);
Rtot = RT + R_added;
%-------------------- MODULE 4: Propulsive efficiency --------------------%
% Wageningen-B series by Breslin and Andersen (1994) and updated

% according to Kristensen (2017)
[etaD_cw] = wageningen(V,RT,etaH,t,w,Dp,rho); % Calm water

[etaD] = wageningen(V,Rtot,etaH,t,w,Dp,rho); % Total
%------------------------- MODULE 5: Final power -------------------------%
% Final propulsion power
PB_cw = ((1./etaD_cw).*(RT).*V)/1000; % Calm water power

PB = ((1./etaD).*(Rtot).*V)/1000; % Total power
XIX
Estimate missing input
function [Lwl,Los,Lfn,TA,TF,voldisp,S,SB,Sapp,AVT,AVL,CB,CM,CP,CWP] ...

= estimate_hull_char(B,T,D,no_decks,teu,dwt,ldt,Lpp,ship_type,rho)
%------------------------------------------------------------------------%
% estimate_hull_char.m - Estimate missing hull parameter input
%
% Other m-files required: main.m
% Subfunctions: none
%
% Inputs:
% T - Draught
% dwt - Dead weight
% rho - Sea water density 1025 kg/m^3
%
% Outputs:
% Lwl - Length of waterline
% Los - Length of surface
% TA - Molded draught aft perpendicular
% TF - Molded draught fore perpendicular
% voldisp - Volume displacement
% S - Wetted surface area
% SB - Wetted base/transom area
% Sapp - Wetted surface area of appendages
% AVT - Area of ship and cargo above waterline (transverse)
% AVL - Area of ship and cargo above waterline (longitudinal)
% ABT - Transverse cross section area of bulb
% CB - Block coefficient
% CM - Midship section coefficient
% CP - Prismatic coefficient
% CWP - Waterplane area coefficient
% Cstern - Shape parameter in Holtrop-Mennen
% LCB - Longitudinal centre of buoyancy
%
% Author: Tone Dale
% April 2020; Last revision: 19-May-2020
%------------------------------------------------------------------------%
% Approximated parameters calculated directly by design parameters
% Lwl from Kristensen (2013) and Tillig (2020)
if contains(ship_type,'container')
Lwl = 1.01*Lpp;
else
Lwl = 1.02*Lpp;
end
% Los from TMR4220 Naval Hydrodynamics matlab code

Los = 1.04*Lwl;
% Calculation of 'Froude length', Lfn for Hollenbach (from TMR4220):

if Los/Lpp < 1
Lfn = Los;
elseif (Los/Lpp >= 1) & (Los/Lpp < 1.1)
Lfn = Lpp+2/3*(Los-Lpp);
elseif Los/Lpp >= 1.1
Lfn = 1.0667*Lpp;
end
% TA and TF assumed to be equal to T
XX
TA = T;
TF = T;
% Volume displacement [m^3]

voldisp = ((ldt+dwt)*1000/rho);
% Block coefficient w.r.t Lpp, breadth moulded and max draught

CB = voldisp/(Lpp*B*T);
if CB > 0.9
disp('Error: CB > 0.9')
end
% Wetted surface area by Mumfords formula updated by Kristensen (2013)

if contains(ship_type,{'bulk dry', 'oil','other bulk'})
S = 0.99*(voldisp/T + 1.9*Lwl*T);
elseif contains(ship_type,'container')
S = 0.995*(voldisp/T + 1.9*Lwl*T);
else
S = 1.025*Lpp*(CB*B + 1.7*T);
end
% Wetted base/transom area

SB = 0;
% Wetted surface area of appendages

Sapp = 0;
% Area of ship and cargo above waterline (transverse and longitudinal)

% According to Kristensen (2017) for tankers/bulk/container
if teu <= 3000
h = interp1(1:300:3000,11:1:20.6,teu);
elseif (teu > 3000) & (teu <= 6000)
h = 24.2;
elseif (teu > 6000) & (teu <= 16000)
h = interp1(6000:2000:16000,24.2:0.5:26.8,teu);
else
disp(' Estimate transverse cross section area error. TEU > 16000')
end
else
h = 3*no_decks + 2; % 2 m for equipment
end
AVT = B*(D-T+h);
AVL = Lpp*(D-T+h);
% Midship section coefficient by Schneekluth and Bertram (1998) Tillig 2020

% NB! CM > 1 for CB >0.9
CM = 0.93 + 0.08*CB;
% Prismatic coefficient
CP = CB/CM;
% Waterplane area coefficient by Bertram and Wobig (Tillig 2020)

% Found in Birk p. 613
% General formula from Papanikolao Ship design: methodologies of
% preliminary design (book) (2014)
if contains(ship_type,{'bulk dry', 'oil','other bulk', 'cargo'})
CWP = 0.763*(CP + 0.34);
CWP = 3.226*(CP - 0.36);
else
CWP = (1+2*CB)/3;
end %if
if CWP > 1
disp('Error: CWP > 1. Set to 0.9')
CWP = 0.9;
end
XXI
end
function [Dp] = estimate_prop_char(T,ship_type)

%------------------------------------------------------------------------%
% estimate_prop_char.m - Estimate missing propeller characteristics
%
%
% Inputs:
% T - Draught (max)
%
% Outputs:
% Dp - Propeller diameter
%
% Author: Tone Dale
%------------------------------------------------------------------------%
Dp = 0.395*T + 1.30; % Kristensen 2017
Dp = 0.623*T - 0.16; % Kristensen 2017
elseif contains(ship_type, 'ro-ro')
Dp = 0.713*T - 0.08; % Kristensen 2017
else
Dp = 0.7*T; % (MAN)
end
end
Calm-water methods requirements
function [GH_req,HM_req,HB_req] = requirements(Lpp,Lwl,Los,B,T,TA,CB,CP,Dp,voldisp,bulb)

%------------------------------------------------------------------------%
% requirements.m - Check if input fulfills calm-water methods requirement
%
%
% Inputs:
% Lwl - Waterline length
% Los - Length over surface
% Lfn - Computational Froude length
% T - Draught
% g - Constant of gravity
% bulb - yes or no
%
% Outputs:
% GH_req - Guldhammer-Harvald dimensions requirement
% HM_req - Holtrop-Mennen dimensions requirement
% HB_req - Hollenbach dimensions requirement
%
% Author: Tone Dale
%------------------------------------------------------------------------%
XXII
% Guldhammer-Harvald requirements
% Define computational length, L

if contains(bulb, 'yes')
L = Los;
elseif contains(bulb, 'no')
L = Lwl;
end %if
M = L/(voldisp^(1/3)); % Length-displacement ratio or slenderness
if (L/B >= 5.0) & (L/B < 8.0) & (CB >= 0.55) & (CB < 0.85)...
& (M >= 4.0) & (M < 6.0)
GH_req = 1;
else
GH_req = 0;
disp('Guldhammer-Harvald requirement not fulfilled')
end
% Holtrop-Mennen requirements
% CP based on Lpp not Lwl

CP = (Lpp/Lwl)*CP;
if (CP >= 0.55) & (CP < 0.85) & (Lwl/B >= 3.9) & (Lwl/B < 9.5)
HM_req = 1;
else
HM_req = 0;
disp('Holtrop-Mennen requirement not fulfilled')
end
% Hollenbach requirements
if (Lpp/B >= 4.71) & (Lpp/B < 7.11) & ... % Length - beam ratio
(B/T >= 1.99) & (B/T < 4) & ... % Breadth - draught ratio
(Dp/TA >= 0.43) & (Dp/TA < 0.84) & ... % Dp - draught ratio
(CB >= 0.49) & (CB < 0.83) & ... % Block coefficient
(Lpp/((voldisp)^(1/3)) >= 4.49) & ... % Length - disp ratio
(Lpp/((voldisp)^(1/3)) < 6.01)
HB_req = 1;
else
HB_req = 0;
disp('Hollenbach requirement not fulfilled')
end % if
end
Guldhammer-Harvald
function [RT,etaH,t,w] = GH(V,Vd,voldisp,Lwl,Lpp,Los,B,T,S,AVT,CB,CP,Dp,...

ship_type,sub_type,teu,bulb,rho,g,nu)
%------------------------------------------------------------------------%
% GH.m - Calculate calm water resistance with new Guldhammer-Harvald method
%
% Subfunctions: air_resistance.m
%
% Inputs:
% V - Ship actual speed
% Vd - Ship design speed
XXIII
% T - Draught
% rho - Seawater density
% nu - Viscosity
% teu - No of containers (in twenty-foot equivalent units)
% ship type - Ship type
% sub_type - Sub type for container ships
% bulb - Either 'yes' or 'no'
%
% Outputs:
% CR - Residual resistance coefficient
% RT - Total calm water resistance
% etaH - Hull efficiency
% t - Thrust deduction factor
% w - Wake fraction
%
% Author: Tone Dale
% April 2020; Last revision: 16-April-2020:
%------------------------------------------------------------------------%
%Revised Harvald (1983) method defined in accordance with Kristensen (2017)

weightdisp = rho*voldisp/1000; % In tons
% Form factor is excluded in the CF part but corrected by hull form U/V
Re = V*Lwl./nu;
CF = 0.075./(log10(Re)-2).^2;
% Incremental resistance coefficient

% CA includes the effects of roughness of the ship hull (Kristensen 2017)
CA = (max(-0.1, 0.5*log10(weightdisp)-0.1*(log10(weightdisp)).^2))*10^-3;
% Axial calm water air resistance CAA (kristensen 2017 and skipshydro comp)
CAA = air_resistance(ship_type,sub_type,AVT,S,teu);
% Residuary resistance (Regression analysis of curves from Harvald, 1983)

% CR is f(M,CP,Fn,constants)
% Define computational length, L

if contains(bulb, 'yes')
L = Los;
elseif contains(bulb, 'no')
L = Lwl;
end %if
M = L/(voldisp^(1/3)); % Length-displacement ratio or slenderness

Fn = V./sqrt(g*Lpp); % Froude number according to Guldhammer-Harvald
AM = voldisp/(Lpp*CP); % Midship section area
CP = voldisp/(L*AM); % Prismatic coefficient based on computational L
A0 = 1.35 - 0.23*M + 0.012*M^2;

A1 = 0.0011*M^9.1;
N1 = 2*M - 3.7;
E = (A0 + 1.5*Fn.^1.8 + A1*Fn.^N1).*(0.98 + (25/((M-2)^4))) + ...

((M-5)^4).*((Fn-0.1).^4);
B1 = 7 - 0.09*M^2;
B2 = (5*CP - 2.5)^2;
B3 = (600*(Fn - 0.315).^2 + 1).^1.5;
G = B1*B2./B3;
H = exp(80.*(Fn-(0.04+0.59*CP) - 0.015*(M-5)));
XXIV
K = 180*Fn.^(3.7).*exp(20*CP-16);
CR_diagram = (E + G + H + K).*10^-3; % Valid for Fn <= 0.33;
% Corrections:
% Correction for B/T ratio deviating from 2.5

if (B/T <= 2.4) || (B/T >= 2.6)
dCR_BT = 0.16*(B/T - 2.5)*10^-3;
else
dCR_BT = 0;
end
% Correction for LCB not placed amidships

dCR_LCB = 0; % Suggested neglected by Kristensen 2017 (Birk 2019)
% Correction for shape/hull form according to Birk 2019

% Modern tankers/bulk carriers with large CB often got U-fore body
% Fast container ships often have entrances with V-shaped stations
% Assuming fast is a design speed above 20 knots
if (contains(ship_type,{'bulk dry', 'oil','other bulk'})) & (CB > 0.85)

dCR_form = -0.1*10^-3;
elseif (contains(ship_type, 'container')) & (Vd > 10)
dCR_form = 0.1*10^-3;
else
dCR_form = 0;
end %if
CR_nobulb = CR_diagram + dCR_BT + dCR_LCB + dCR_form;
% Bulb correction as function of Fn, according to Kristensen 2017

% Only presented for tankers/bulk carriers
% Correction for container ships is presented as a fraction
% A factor of 10^-3 is missing in Kristensen 2017

dCR_bulb = (max(-0.4,-0.1-1.6.*Fn)).*10^-3;
elseif contains(ship_type, 'container')
dCR_bulb = ((250.*Fn-90)./100).*CR_nobulb;
else
dCR_bulb = 0;
end % if
CR = CR_diagram + dCR_BT + dCR_LCB + dCR_form + dCR_bulb;
% CT by ITTC method
CT = CF + CA + CAA + CR;
RT = CT.*0.5*rho*S.*V.^2;
% Hull-propeller interaction parameters
% Wake fraction
% Fa = [ -2 0 2 ]; Form in the aft body
Fa = 0; % Temporarily assumption
a = 0.1*B/L + 0.149;
b = 0.05*B/L + 0.449;
c = 585 - 5027*B/L + 11700*(B/L)^2;
w1 = a + b/(c*(0.98-CB)^3+1);
w2 = 0.025*Fa/(100*(CB-0.7)^2+1);
w3 = min(0.1,-0.18 + 0.00756/(Dp/L + 0.002));
w_harvald = w1 + w2 + w3;
w = (0.7*w_harvald - 0.45 + 0.08*M).*ones(1,length(V));
% Thrust deduction factor

d = 0.625*B/L + 0.08;
e = 0.165 - 0.25*B/L;
f = 825 - 8060*B/L + 20300*(B/L)^2;
XXV
t1 = d + e/(f*(0.98-CB)^3+1);
t2 = -0.01*Fa;
t3 = 2*(Dp/L - 0.04);
t_harvald = t1 + t2 + t3;
t = (t_harvald - 0.26 + 0.04*M).*ones(1,length(V));
etaH = (1-t)./(1-w);
% Froude number limitations

RT(Fn > 0.33) = 0;
w(Fn > 0.33) = 0;
t(Fn > 0.33) = 0;
etaH(Fn > 0.33) = 0;
end
Holtrop-Mennen
function [RT,etaH,t,w] = HM(V,Vd,voldisp,Lwl,Lpp,B,TA,TF,S,Sapp,CB,CP,...

CM,CWP,Dp,rho,g,nu,ship_type,sub_type,AVT,teu)
%------------------------------------------------------------------------%
% HM.m - Calculate calm water resistance with Holtrop-Mennen
%
%
% Inputs:
% Vd - Ship design speed
% T - Draught
% Sapp - Wetted surface area of appendages (= 0)
% CWP - Waterplane area coefficient
% nu - Viscosity
% sub_type - Sub type for container ships
%
% Outputs:
% w - Wake fraction
%
% Author: Tone Dale
% April 2020; Last revision: 16-June-2020:
%------------------------------------------------------------------------%
% Method implemented as described by Birk 2019 p.613-616
Fn = V./sqrt(g*Lwl);
Fn_design = Vd/sqrt(g*Lwl);
Re = V*Lwl./nu;
XXVI
% CP and CB defined with Lwl
CP = (Lpp/Lwl)*CP;
CB = (Lpp/Lwl)*CB;
T = (TA+TF)/2;
% Transverse area of bulbous bow

% Kracht 1978 gives C_ABT (Tillig 2020)
C_ABT = (40*Fn_design-3.5)/100;
AMS = B*T*CM;
ABT = C_ABT*AMS;
%Height of centre of bulb area max 0.6*TF according to Birk 2019 p. 612
hB = 0.6*TF;
% Assumption: Area of immersed transom = 0 birk p.623

AT = 0;
% Optimum LCB: lcb estimated by Guldhammer Harvald (Birk p.613)

% [%] aft of Lwl/2
lcb = -(0.44*Fn_design - 0.094);
LR = Lwl*(1-CP + (0.06*CP*lcb)/(4*CP-1)); % Length of run
% Waterline entrance angle iE [degrees]

a = -( ((Lwl/B)^0.80856)*((1-CWP)^0.30484)*((1-CP-0.0225*lcb)^0.6367)*...
((LR/B)^0.34574)*((100*voldisp/(Lwl^3))^0.16302) );
iE = 1 + 89*exp(a);
Cstern = 0; % Assuming normal sections

c14 = 1.0 + 0.011*Cstern; % Constant of stern shape
% Hull form factor

k = -0.07 + 0.487118*c14*(((B/Lwl)^1.06806)*((T/Lwl)^0.46106)*...
((Lwl/LR)^0.121563)*((Lwl^3/voldisp)^0.36486)*(1-CP)^-0.604247);
% Frictional resistance by ITTC 1957 line

CF = 0.075./(log10(Re)-2).^2;
RF = CF.*0.5*rho*S.*V.^2;
%% % Wave resistance defined according to Froude number range
if B/Lwl <= 0.11

c7 = 0.229577*(B/Lwl)^(1/3);
elseif (B/Lwl > 0.11) & (B/Lwl <= 0.25)
c7 = B/Lwl;
else
c7 = 0.5-0.0625*Lwl/B;
end
c1 = 2223105*(c7^3.78613)*((T/B)^1.07961)*(90-iE)^-1.37565;
c3 = 0.56*(ABT^1.5)/(B*T*(0.31*sqrt(ABT) + TF - hB));
c2 = exp(-1.89*sqrt(c3));
c5 = 1-0.8*AT/(B*T*CM);
if (Lwl^3)/voldisp <= 512

c15 = -1.69385;
elseif ((Lwl^3)/voldisp > 512) & ((Lwl^3)/voldisp <= 1726.91)
c15 = -1.69385 + ((Lwl/(voldisp^(1/3))) - 8)/2.36;
else
c15 = 0;
end
if CP <= 0.8
c16 = 8.07981*CP - 13.8673*CP^2 + 6.984388*CP^3;
else
c16 = 1.73014 - 0.7067*CP;
end
XXVII
d = -0.9;
if Lwl/B <= 12
lambda = 1.446*CP - 0.03*Lwl/B;
else
lambda = 1.446*CP - 0.36;
end
m1 = 0.0140407*Lwl/T - 1.75254*(voldisp^(1/3))/Lwl - 4.79323*B/Lwl - c16;
m4 = 0.4*c15.*exp(-0.034*Fn.^-3.29);
RW = c1*c2*c5*rho*g*voldisp.*exp(m1.*Fn.^d + m4.*cos(lambda.*Fn.^-2));
%% Other resistance contributions

% Neglecting effect of appendage resistance
RAPP = 0;
% Neglecting the bulb correction

RB = 0;
% Neglecting transom resistance

RTR = 0;
% Correlation coefficient includes roughness
if TF/Lwl <= 0.04

c4 = TF/Lwl;
else
c4 = 0.04;
end
CA = 0.006*((Lwl+100)^-0.16) - 0.00205 + 0.003*sqrt(Lwl/7.5)*(CB^4)*c2*(0.04-c4);
% Additional correction for surface roughness > 150 um

ks = 150; % Assumption
if ks == 150
dCA = 0;
else
dCA = (0.105*(ks*10^-6)^(1/3) - 0.005579)/(Lwl^(1/3));
end
% Correlation resistance
RA = 0.5*rho.*(V.^2)*(CA+dCA)*(S+Sapp);
% Air resistance
RAA = 0.5*rho*S.*(V.^2)*CAA;
%% Total Resistance
RT = (1+k).*RF + RW + RAA + RA + RAPP + RB + RTR;
%% Hull-propeller interation parameters
% Viscous resistance coefficient

CV = ((1+k).*RF + RA + RAPP)./(0.5*rho.*(V.^2)*(S+Sapp));
% Wake fraction
if B/TA <= 5
c8 = (S/(Lwl*Dp))*(B/TA);
else
c8 = (S*(7*B/TA - 25))/(Lwl*Dp*(B/TA - 3));
end
if c8 <= 28
c9 = c8;
XXVIII
else
c9 = 32 - 16/(c8-24);
end
if TA/Dp <= 2
c11 = TA/Dp;
else
c11 = 0.0833333*((TA/Dp)^3) + 1.33333;
end
if CP <= 0.7
c19 = 0.12997/(0.95-CB) - 0.11056/(0.95-CP);
else
c19 = 0.18567/(1.3571-CM) - 0.71276 + 0.38648*CP;
end
c20 = 1+0.015*Cstern;
CP1 = 1.45*CP - 0.315 - 0.0225*lcb;
w = c9*c20*CV.*(Lwl/TA).*(0.050776 + 0.93405*(c11.*CV)./(1-CP1)) + ...

0.27915*c20*sqrt(B/(Lwl*(1-CP1))) + c19*c20; % ! Array
% Thrust deduction factor

t = (0.25014*((B/Lwl)^0.28956)*(((sqrt(B*T))/Dp)^0.2624))/((1-CP+0.0225*lcb)^0.01762) + 0.0015*Cstern;
t = t.*ones(1,length(V));
% Hull efficiency
etaH = (1-t)./(1-w);
%% Froude number limitations

RT(Fn > 0.45) = 0;
w(Fn > 0.45) = 0;
t(Fn > 0.45) = 0;
etaH(Fn > 0.45) = 0;
end
Hollenbach
function [RT_mean,etaH,t,w] = hollenbach(V,Lwl,Lpp,Los,Lfn,B,TA,TF,S,CB,Dp,...

rho,g,nu,ship_type,sub_type,AVT,teu)
%------------------------------------------------------------------------%
% hollenbach.m - Calculate calm water resistance with Hollenbach (mean)
%
%
% Inputs:
% V - Ship speed
% Lwl - Length in waterline
% Lfn - Computational Froude length
% T - Draught
% TA - Molded draught fore perpendicular
% nu - Viscosity
%
% Outputs:
% RT_mean - Total calm water resistance with k, dF and CA (mean)
XXIX
% w - Wake fraction
%
% Author: Tone Dale
% April 2020; Last revision: 16-May-2020:
%------------------------------------------------------------------------%
% Script based on Skipshydro Hollenbach script for mean resistance (credit)
NRud = 0; % Number of rudders

NBrac = 0; % Number of brackets
NBoss = 0; % Number of bosses
NThr = 0; % Number of side thrusters
T = (TA+TF)/2;
N = length(V);
% 'Mean' resistance coefficients

a = [-0.3382 0.8086 -6.0258 -3.5632 9.4405 0.0146 0 0 0 0];
b = [-0.57424 13.3893 90.5960; 4.6614 -39.721 -351.483; -1.14215...
-12.3296 459.254];
d = [0.854 -1.228 0.497];
e = [2.1701 -0.1602];
% Correction for ships with CB<0.6 by Hollenbach 1998 (Birk 2019)

if CB < 0.49
b(1) = -0.87674;
elseif (CB >= 0.49) & (CB < 0.6)
b(1) = -0.57424-25*(0.6-CB)^2;
else
b(1) = -0.57424;
end
CRstandard = zeros(1,N); % Pre-allocation
% Form factor by MARINTEK, eq. 2.54 in Skipsydro comp

Q = (CB/Lwl)*sqrt(B*(TA+TF));
k = 0.6*Q + 75*Q^3;
% Calculate ship resistance
Fn = V./sqrt(g*Lfn); % Froude number according to Hollenbach

Fnkrit = d*[1 CB CB^2]';
c1 = Fn./Fnkrit;
% Other resistance components

Re = V*Lpp./nu; % Reynold's number for ship
CF = 0.075./(log10(Re)-2).^2; % Friction coefficient for ship
Rnm = (6*sqrt(6/Lpp)*10^6/1.1395).*V; % Reynold's number for model
CFm = 0.075./(log10(Rnm)-2).^2; % ITTC friction line for model
dCF = (110.31.*((150*V).^0.21) - 403.33).*(CF.^2); % MARINTEK roughness
dCF(dCF<0) = 0; % Only positive values
% CA from Vienna Model Basin p.637 Birk

if Lpp <= 175
CA = (0.35-0.002*Lpp)/1000;
else
CA = 0;
end
% Axial calm water air resistance CAA

XXX
% Residual resistance component
CRFnkrit = max(1.0,(Fn./Fnkrit).^c1);
kL = e(1)*Lpp^(e(2));
for i = 1:N
CRstandard(i) = [1 CB CB^2]*(b*[1 Fn(i) Fn(i)^2]')/10;
end
CR_hollenbach = CRstandard.*CRFnkrit*kL*prod([T/B B/Lpp Los/Lwl...

Lwl/Lpp (1+(TA-TF)/Lpp) Dp/TA (1+NRud) (1+NBrac) (1+NBoss)...
(1+NThr)].^a);
% When accounting for form factor

CR_mean = CR_hollenbach.*B*T/S - k.*CFm;
CT_mean = (1+k).*(CF + dCF) + CR_mean + CA + CAA;
RT_mean = CT_mean.*0.5*rho*S.*V.^2;
% Power prediction
% According to Birk, Hollenbach uses a fixed value of t = 0.19 for single
% screw ships and 0.15 for twin screw
% The wake fraction is computed by the model hull efficiency (Birk p.641)
t = 0.19*ones(1,length(V));
etaH_model = 0.948*CB^0.3977*((RT_mean/RT_mean)^(-0.58))*...
((B/T)^0.1727)*(Dp^2/(B*T))^-0.1334;
w_model = (1 - (1-t)./etaH_model).*ones(1,length(V));
w = (t+0.04) + (w_model-(t+0.04)).*((CF+CA)/CFm);
etaH = (1-t)./(1-w);
% Froude number limitation

f_HB = [0.17 0.20 0.60]; %'Mean' range for Hollenbach
g_HB = [0.642 -0.635 0.150]; %'Mean' range for Hollenbach
Fn_min = min(f_HB(1),f_HB(1)+f_HB(2)*(f_HB(3)-CB));
Fn_max = g_HB(1) + g_HB(2)*CB + g_HB(3)*CB^2;
RT_mean(Fn < Fn_min) = 0;

RT_mean(Fn > Fn_max) = 0;
w(Fn < Fn_min) = 0;

w(Fn > Fn_max) = 0;
etaH(Fn < Fn_min) = 0;

etaH(Fn > Fn_max) = 0;
end
STAWAVE-1
function Rwave = STA1(rho,g,Hs,wave_angle,B,Lpp)

%------------------------------------------------------------------------%
% STA1.m - Calculate added resistance in waves with STAWAVE-1 by MARIN
%
%
% Inputs:
% rho - Sea water density
% wave_angle - Wave angle relative to ship heading direction
% B - Breadth
% L_bwl - Waterline length to 95% of maximum breadth
%
XXXI
% Outputs:
% Rwave - Added resistance in waves
%
% Author: Tone Dale
%------------------------------------------------------------------------%
angle = abs(wave_angle); % Wave angle defined for [0,+-180]
L_bwl = 0.18*Lpp; % Comparison ship for Case 1
Rwave= (1/16)*rho*g*Hs.^2*B*sqrt(B/L_bwl);
% Restrictions for wave height and wave direction:

Rwave(Hs > 2.25*sqrt(Lpp/100)) = 0;
Rwave(angle > 45) = 0;
end
STAWAVE-2
function [R_AWL,STA2_req] = STA2(rho,g,Hs,H,B,Lwl,Lpp,Twave,V,...

CB,T,wave_angle)
%------------------------------------------------------------------------%
% STA2.m - Calculate added resistance in waves with STAWAVE-2 by MARIN
%
%
% Inputs:
% H - Mean wave height
% B - Breadth
% Twave - Mean wave period
% T - Draught
% U - Wind speed relative to ship
%
% Outputs:
% R_AWL - Added resistance in waves based on PM wave spectrum
% STA2_req - Logical value, 1 if requirement is fulfilled, else 0
%
% Author: Tone Dale
%------------------------------------------------------------------------%
% U is U_true
% Parameters for each timestep

Z = H/2; % Wave amplitude
T1 = Twave;
R_AWL = zeros(1,length(V));
% Wave sprectrum is required. As recommended by MARIN when only Hs and

% Twave is known, the spectrum is approximated by a theoretical shape.
% Applying the ISSC (modified PM) formula
XXXII
% Seastate description, constant parameters
omega = 2*pi*(0.01:0.005:0.5); % Circular frequency of regular waves
dw = (max(omega)-min(omega))/length(omega);
k = (omega.^2)./g; % Deep water dispersion relation
kyy = 0.25; % Radius of gyration in lateral direction according to ITTC
%% Loop over timesteps and compute R_AWL with respective Spectrum S(w)
for i = 1:length(Fn)
W = ((sqrt(Lpp/g)*(kyy)^(1/3))./(1.17.*Fn(i)^-0.143)).*omega;
a1 = 60.3*CB^1.34.*ones(1,length(omega));
b1 = -8.5*ones(1,length(omega));
b1(W < 1) = 11;
d1 = -566*(Lpp/B)^(-2.66)*ones(1,length(omega));
d1(W < 1) = 14;
raw = (W.^b1).*exp((b1./d1).*(1-W.^d1)).*a1.*Fn(i)^(1.5).*exp(-3.5.*Fn(i));
R_AWML = 4*rho*g*Z(i)^2*((B^2)/Lpp).*raw;
I1 = besseli(1,1.5*k*T); % modified Bessel function of the first kind
K1 = besselk(1,1.5*k*T); % modified Bessel function of the second kind
f1 = 0.692.*(V(i)/sqrt(T*g)).^0.769 + 1.81*CB^6.95;
alpha = ((pi^2*I1.^2)./(pi^2*I1.^2 + K1.^2)).*f1;
R_AWRL = 0.5*rho*g*Z(i).^2*B.*alpha;
Rwave = R_AWML + R_AWRL;
% ISSC Spectrum
S = (173*((Hs(i)^2)/(T1(i)^4)))*omega.^(-5).*exp((-692./(T1(i).^4)).*...
omega.^(-4));
Aj_squared = 2.*S.*dw;
Fj = (Rwave/(Z(i)^2)).*Aj_squared;
R_AWL(i) = sum(Fj);
end
%% Restrictions
R_AWL(Fn > 0.30) = 0;
R_AWL(Fn < 0.1) = 0;
angle = abs(wave_angle); % Wave angle defined for [0,+-180]

R_AWL(angle > 45) = 0;
if (Lpp > 75) & (Lpp/B >= 4.0) & (Lpp/B < 9.0) & (B/T >= 2.2) & ...
(B/T < 5.5) & (CB >= 0.5) & (CB < 0.9)
STA2_req = 1;
else
STA2_req = 0;
disp('STAWAVE-2 requirement not fulfilled')
end % if
end
STAJIP-wind
function RAA = STAJIP_wind(ship_type,sub_type,AVT,rho_air,U,U_angle,V)
XXXIII
%------------------------------------------------------------------------%
% STAJIP_wind.m - Calculate added resistance in wind by MARIN
%
%
% Inputs:
% ship_type - Ship type
% sub_type - Sub ship type
% AVT - Transverse projected area of superstructure
% rho_air - Air density
% V - Ship speed over ground
%
% Outputs:
% RAA - Added resistance in wind by STAJIP
%
% Author: Tone Dale
%------------------------------------------------------------------------%
% Here V = Vsog
Urel = U;
%% Drag coefficients read off STA-JIP plots in ISO 15016

% Tanker, laden, conventional bow (Assuming: oil, bulk dry, other bulk)
Cx_tanker = [-0.75 -0.77 -0.72 -0.64 -0.51 -0.39 -0.28 -0.20 -0.12 0.03...
0.09 0.19 0.34 0.5 0.61 0.74 0.86 0.93 0.98];
% Containership, design condition, with containers

Cx_container = [-0.68 -0.73 -0.74 -0.68 -0.49 -0.32 -0.26 -0.21 -0.22...
-0.27 -0.14 0.1 0.36 0.60 0.77 0.89 0.89 0.80 0.66];
% Passenger/Cruise
Cx_passenger = [-0.70 -0.72 -0.73 -0.70 -0.48 -0.24 -0.26 -0.1 0.09 0.05...
-0.05 0.09 0.22 0.38 0.57 0.72 0.80 0.71 0.66];
% General cargo
Cx_cargo = [-0.6 -0.89 -1.0 -1.0 -0.89 -0.84 -0.65 -0.43 -0.28 -0.1 0.09...
0.49 0.83 1.11 1.39 1.49 1.33 0.91 0.81];
% Large LNG carrier. Average of prismatic and spherical for following sea
Cx_LNG = [-1.01 -0.99 -0.92 -0.81 -0.67 -0.49 -0.30 -0.15 -0.04 0.03 0.10...
0.22 0.42 0.65 0.78 0.88 0.95 0.96 0.94];
if contains(ship_type,{'bulk dry', 'oil', 'other bulk'})

CDA_direction = -Cx_tanker;
CDA_direction = -Cx_container;
elseif contains(ship_type,'passenger')
CDA_direction = -Cx_passenger;
elseif contains(ship_type,'cargo')
CDA_direction = -Cx_cargo;
elseif contains(sub_type,'LNG')
CDA_direction = -Cx_LNG;
else
CDA_direction = 0;
disp('STA-JIP wind method not applicable');
end
%% Assign correct CAA value based on actual wind direction in each timestep
x = [0 10:10:180];
U_range = U_angle; % Transform 0 to 360 degrees to between 0 and 180
U_range(U_range>180) = abs(U_range(U_range>180)-360);
CDA = interp1(x,CDA_direction,U_range);
RAA = 0.5*rho_air*CDA*AVT.*Urel.^2 - 0.5*rho_air*CDA_direction(1)*AVT*V.^2;
XXXIV
end
Blendermann wind
function RAA = blendermann_wind(ship_type,sub_type,AVT,rho_air,U,U_angle,V)

%------------------------------------------------------------------------%
% blendermann_wind.m - Compute wind resistance accoring to Blendermann
%
%
% Inputs:
% sub_type - Sub types defined for tankers/bulk vessels
% V - Ship speed over ground
%
% Outputs:
% RAA - Wind resistance
%
% Author: Tone Dale
% April 2020; Last revision: 15-July-2020:
%------------------------------------------------------------------------%
% Here V = Vsog
Urel = U;
%% Drag coefficients read off Blendermann diagrams from Brix 1993
% Car carrier
Cx_carcarrier = [-0.55 -0.65 -0.65 -0.60 -0.40 -0.10 0.18 0.15 0 -0.20...
-0.30 -0.25 -0.12 0.20 0.50 0.70 0.80 0.74 0.60];
% Containership
Cx_container = [-0.55 -0.47 -0.46 -0.52 -0.52 -0.44 -0.34 -0.26 -0.11...
-0.03 0.08 0.22 0.44 0.58 0.60 0.63 0.58 0.54 0.40];
% Tanker, loaded
Cx_tanker = [-0.90 -0.87 -0.95 -0.95 -0.85 -0.72 -0.51 -0.29 -0.03 0.10...
0.26 0.32 0.42 0.51 0.57 0.62 0.61 0.60 0.55];
% General cargo
Cx_cargo = [-0.65 -0.67 -0.77 -0.77 -0.70 -0.52 -0.37 -0.21 -0.05 0.15...
0.22 0.29 0.29 0.37 0.47 0.57 0.65 0.59 0.60];
% Passenger
Cx_passenger = [-0.40 -0.37 -0.35 -0.27 -0.08 0.09 0.04 0 -0.01 -0.03...
-0.08 -0.08 -0.16 -0.18 -0.06 0.21 0.33 0.38 0.37];
% Offshore Supply ship

Cx_offshore_supply = [-0.55 -0.55 -0.51 -0.50 -0.43 -0.33 -0.25 -0.17...
-0.12 -0.10 -0.07 0.07 0.37 0.59 0.71 0.79 0.75 0.81 0.81];
% Deep sea drilling ship

Cx_drilling = [-1.05 -1.22 -1.39 -1.38 -1.27 -1.08 -0.78 -0.56 -0.34...
-0.15 -0.04 0.23 0.67 1.01 1.23 1.37 1.34 1.25 1.25];
% Ro-ro/Lo-lo
Cx_roro = [-0.66 -0.66 -0.74 -0.71 -0.67 -0.55 -0.40 -0.25 -0.14 -0.09...
-0.15 -0.05 0 0.26 0.52 0.81 0.88 0.79 0.70];
% Research vessel
XXXV
Cx_research = [-0.55 -0.59 -0.64 -0.59 -0.54 -0.47 -0.40 -0.29 -0.21...
-0.18 -0.15 -0.06 0.10 0.39 0.62 0.82 0.83 0.71 0.60];
if contains(ship_type,'car carrier')
CDA_direction = -Cx_carcarrier;
elseif contains(ship_type,{'bulk dry', 'oil', 'other bulk'})
CDA_direction = -Cx_tanker;
CDA_direction = -Cx_container;
elseif contains(ship_type,'passenger')
CDA_direction = -Cx_passenger;
elseif contains(ship_type,'cargo')
CDA_direction = -Cx_cargo;
elseif contains(ship_type, 'offshore supply')
CDA_direction = -Cx_offshore_supply;
elseif contains(ship_type,'deep sea drilling')
CDA_direction = -Cx_drilling;
elseif contains(sub_type,'ro-ro')
CDA_direction = -Cx_roro;
elseif contains(sub_type,'research')
CDA_direction = -Cx_research;
else
CDA_direction = 0;
disp('Blendermann wind method not applicable');
end
%% Assign correct CAA value based on actual wind direction in each timestep
x = [0 10:10:180];
U_range = U_angle; % Transfort 0 to 360 degrees to between 0 and 180
U_range(U_range>180) = abs(U_range(U_range>180)-360);
CDA = interp1(x,CDA_direction,U_range);
RAA = 0.5*rho_air*CDA*AVT.*Urel.^2 - 0.5*rho_air*CDA_direction(1)*AVT*V.^2;
end
Townsin & Kwon
function Radded = kwon(CB,Lwl,V,H,wave_angle,U,U_angle,g,voldisp,ship_type,RT)

%------------------------------------------------------------------------%
% kwon.m - Calculate speed reduction due to added resistance in wind and
% waves with the Townsin and Kwon method
%
% Subfunctions: beaufort.m
%
% Inputs:
% V - Ship speed
% Twave - Mean wave period
% U - Wind speed in ship heading direction
% U_angle - Wind speed direction relative to ship heading dir.
% RT - Calm water resistance
%
% Outputs:
% Radded - Added resistance in wind and waves
%
% Author: Tone Dale
XXXVI
% April 2020; Last revision: 21-June-2020:
%------------------------------------------------------------------------%
% U is U_true
% Rounding CB to nearest value defined by Kwon

CB = (round(100*CB/5)*5)/100;
% Speed reduction CU due to block coefficient

CU = zeros(1,length(Fn));
if CB == 0.55
CU = 1.7 - 1.4*Fn - 7.4*Fn.^2;
elseif CB == 0.60
CU = 2.2 - 2.5*Fn - 9.7*Fn.^2;
elseif CB == 0.65
CU = 2.6 - 3.7*Fn - 11.6*Fn.^2;
elseif CB == 0.70
CU = 3.1 - 5.3*Fn - 12.4*Fn.^2;
elseif CB == 0.75
CU = 2.4 - 10.6*Fn - 9.5*Fn.^2;
elseif CB == 0.80
CU = 2.6 - 13.1*Fn - 15.1*Fn.^2;
elseif CB == 0.85
CU = 3.1 - 18.7*Fn + 28*Fn.^2;
end % if
% Direction reduction coefficient Cbeta due to weather direction

% wave_angle [0,+-180] and U_angle [0,360] relative to the bow
[BN_wave,BN_wind] = beaufort(H,U);
% Applying BN according to wind definition scale

BN = BN_wind;
angle = U_angle;
% For angle defined by waves, use absolute value instead:

Cbeta = zeros(1,length(Fn));
head_sea = ((angle >= 0) & (angle < 30)) | ((angle >= 330) & (angle < 360));
bow_sea = ((angle >= 30) & (angle < 60)) | ((angle >= 300) & (angle < 330));
beam_sea = ((angle >= 60) & (angle < 150)) | ((angle >= 210) & (angle < 300));
following_sea = ((angle >= 150) & (angle < 210));
Cbeta(head_sea)= 1;
Cbeta(bow_sea)= 0.5*(1.7-0.03*(BN(bow_sea)-4).^2);
Cbeta(beam_sea)= 0.5*(0.9-0.06*(BN(beam_sea)-6).^2);
Cbeta(following_sea)= 0.5*(0.4-0.03*(BN(following_sea)-8).^2);
% Form coefficient - Sea-web data is generally for loaded condition
Cform = 0.7*BN + (BN.^6.5)/(22*voldisp^(2/3));
else
Cform = 0.5*BN + (BN.^6.5)/(2.7*voldisp^(2/3));
end
x = (CU.*Cbeta.*Cform)./100;
Radded = x.*RT;
% Restricted to Froude number range below 0.3

Radded(Fn>0.3) = 0;
end
Wageningen B-series
function [etaD] = wageningen(V,RT,etaH,t,w,Dp,rho)

%------------------------------------------------------------------------%
XXXVII
% wageningen.m - Calculate open water efficiency by Wageningen B-series,
% Approximated method by Breslin and Andersen (Kristensen 2017)
%
%
% Inputs:
% V - Ship speed
% w - Wake fraction
%
% Outputs:
% etaD - Propulsive efficiency
%
% Author: Tone Dale
% April 2020; Last revision: 21-April-2020:
%------------------------------------------------------------------------%
VA = (1-w).*V;
CTh = (8/pi)*RT./((1-t).*rho.*(VA.*Dp).^2);
eta0_ideal = 2./(1+sqrt(CTh+1));
eta0 = eta0_ideal.*(max(0.69,0.81-0.014.*CTh));
% Relative rotative efficiency etaR = 1.0 assumed

etaR = 1.0;
% Shaft efficiency etaS = 0.98 for direct drive and 0.97-0.96 gearbox
% (Kristensen 2017)
etaS = 0.98;
% Final propulsive efficiency

etaD = eta0.*etaH*etaR*etaS;
end
Beauforts scale
function [BN_wave,BN_wind] = beaufort(H,U)

%------------------------------------------------------------------------%
% beafort.m - Assign beauforts number to weather condition
%
% Other m-files required: none
%
% Inputs:
% H - Significant wave height
% U - True wind speed
%
% Outputs:
% BN_wind - Beauforts number from wind scale
% BN_wave - Beauforts number from wave scale
%
% Author: Tone Dale
%------------------------------------------------------------------------%
% Wave BN from Henschke 1956 (Kwon 2008)

BN = zeros(1,length(H));
BN(H<0) = nan;
XXXVIII
BN((H >= 0) & (H < 0.5)) = 1;
BN((H >= 0.5) & (H < 0.65)) = 2;
BN((H >= 0.65) & (H < 0.75)) = 3;
BN((H >= 0.75) & (H < 1.25)) = 4;
BN((H >= 1.25) & (H < 2.0)) = 5;
BN((H >= 2.0) & (H < 3.5)) = 6;
BN((H >= 3.5) & (H < 6.0)) = 7;
BN((H >= 6.0) & (H < 8.0)) = 8;
BN(H >= 8.0) = nan;
BN_wave = BN;
BN = zeros(1,length(U));
% Wind Beaufort Scale according to ISO 15016

BN((U >= 0) & (U < 0.3)) = 0;
BN((U >= 0.3) & (U < 1.6)) = 1;
BN((U >= 1.6) & (U < 3.4)) = 2;
BN((U >= 3.4) & (U < 5.5)) = 3;
BN((U >= 5.5) & (U < 8.0)) = 4;
BN((U >= 8.0) & (U < 10.8)) = 5;
BN((U >= 10.8) & (U < 13.9)) = 6;
BN((U >= 13.9) & (U < 17.2)) = 7;
BN((U >= 17.2) & (U < 20.7)) = 8;
BN(U >= 20.7) = nan;
BN_wind = BN;
end
Air resistance
function CAA = air_resistance(ship_type,sub_type,AVT,S,teu)

%------------------------------------------------------------------------%
% air_resistance.m - Calculate air resistance
%
%
% Inputs:
% sub_type - Sub types defined for tankers/bulk vessels
%
% Outputs:
% CAA - Axial calm water air resistance
%
% Author: Tone Dale
%------------------------------------------------------------------------%
% Check ship type and assign CAA
% Containers (Kristensen, 2017)

CAA = (max(0.09,0.28*teu^-0.126))*10^-3;
% Bulk/Tankers (Kristensen, 2017)
elseif contains(sub_type,'small')
CAA = 0.07*10^-3;
elseif contains(sub_type,'handysize')
CAA = 0.07*10^-3;
elseif contains(sub_type,'handymax')
CAA = 0.07*10^-3;
elseif contains(sub_type,'panamax')
CAA = 0.05*10^-3;
XXXIX
elseif contains(sub_type,'aframax')
CAA = 0.05*10^-3;
elseif contains(sub_type,'suezmax')
CAA = 0.05*10^-3;
elseif contains(sub_type,'vlcc')
CAA = 0.04*10^-3;
% Remaining drag coefficient values set to 0.8 based on Blendermann
% and Birk 2019 p. 620
else
CAA = 0.8*AVT/(800*S);
end
end
Global variables
%------------------------------------------------------------------------%
% Variables.m - Global variable definitions
% Note! Local variables are defined and described locally in functions
%
%
% Author: Tone Dale
% April 2020; Last revision: 16-April-2020
%------------------------------------------------------------------------%
%
%------------------------------ Constants -------------------------------%
%
% nu - Sea water kinematic viscosity
%
%------------------------- Hull characteristics -------------------------%
%
% Los - Length of surface
% B - Ship breadth (moulded, not extreme)
% T - Ship draught (maximum druaght summer load line)
% TF - Molded draught fore perpendicular
% SB - Wetted base/transom area
% Sapp - Wetted surface area of appendages
% AVL - Area of ship and cargo above waterline (longitudinal)
% ABT - Transverse cross section area of bulb (H-M)
% Cstern - Shape parameter in Holtrop-Mennen
% LCB - Longitudinal centre of buoyancy (H-M)
% V - Actual speed
% Vd - Design speed
% dwt - Dead weight
%
%----------------------- Propeller characteristics ----------------------%
%
% AEA0 - Propeller expanded ratio
% n - Propeller rpm
XL
%
%----------------------------- Resistance -------------------------------%
%
% CT - Total calm water resistance coefficient
% CR - Residuary resistance coefficients
% k - Form factor
% CF - Frictional resistance coefficients
% dCF - Hull roughness coefficient
% CDB - Base drag coefficient
% CAA - Air resistance coefficient
% Capp - Appendage resistance coefficient
% CA - Correlation allowance
% Fn - Froude number
% Re - Reynolds number
% Rwind - Added resistance due to wind
% Rwave - Added resistance due to waves
% R_AWML - Resistance due to induced wave motions
% R_AWRL - Resistance due to wave reflection
% Rtot - Total resistance (RT+Radded)
% h - Water depth
% ZA - Wave amplitude
% Vw - Reduced speed corrected for added resistance
%
%----------------------------- Propulsion -------------------------------%
%
% etaD - Propulsive efficiency in ideal conditions
% etatot - Propulsive efficiency in trial conditions
% eta0 - Open water (propeller efficiency)
% etaR - Relative rotative efficiency
% etaS - Shaft efficiency
% PD - Deliwered power at the propeller
% w - Wake fraction
% VA - Speed of inflow to the propeller
% T - Torque
% PS - Calm water power requirement
% Ptot - Total power requirement
% zetaP - Slope of linear curve in the load variation test
% ME_n - Main engine rpm
% ME_cyl - Main engine no. of cylinders
% SM - Sea margin
%
%----------------------------- Ship types -------------------------------%
% general_cargo
% bulk_dry
% oil
% passenger
% offshore_supply
% container
% chemical
% ro_ro
% other_offshore
% liquefied_gas
% other_bulk
% refrigerated_cargo
% other_cargo
% other_liquids
XLI

No - Ntnu Inspera 54166542 26312328

Uploaded by

Copyright:

Available Formats

No - Ntnu Inspera 54166542 26312328

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

No - Ntnu Inspera 54166542 26312328

Uploaded by

Copyright:

Available Formats

Tone Dale 2020 Master's thesis

Norwegian University of Science and Technology

MASTER THESIS IN MARINE TECHNOLOGY

Development of simplified methods for ship powering performance

The thesis shall be submitted electronically (pdf) in Inspera:

Supervisor : Professor Sverre Steen

Trondheim, July, 2020

2 Theory of Ship Powering 5

3 Empirical Prediction Methods 15

4 The MariTEAM Model 35

5 New Performance Prediction Method 43

6 Results and Validation 59

9 Recommendations for Further Work 105

Appendix A Hollenbach Coefficients I

Appendix B Guldhammer-Harvald Coefficients III

Appendix C Holtrop-Mennen Coefficients V

Appendix D Sea-web Parameters VII

Appendix E Blendermann Wind Coefficients XI

Appendix F The New Program (MATLAB code) XVII

3.1 Empirical resistance procedures with limited applicable area . . . . . . . . . . . . . . . . . . . 15

4.1 Relevant Sea-web parameters for the power performance predictions . . . . . . . . . . . . . . 37

6.1 Overview of case vessel data in the case study . . . . . . . . . . . . . . . . . . . . . . . . . . 59

E.1 Car carrier wind coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI

2.1 Flow features of a moving ship based on Molland et al. (2017) . . . . . . . . . . . . . . . . . . 5

3.1 Definition of LBW L in STAWAVE-1 (ISO Technical Committee, 2015) . . . . . . . . . . . . . . 20

6.1 In-service routes for the recorded voyage data . . . . . . . . . . . . . . . . . . . . . . . . . . 60

A.1 Coefficients for computing CR in Hollenbach as presented by Birk (2019) . . . . . . . . . . . . I

2.1 Total calm water resistance coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Hull and Propeller Characteristics

1.1 Background and Motivation

1.3 Outline of Thesis

Theory of Ship Powering

2.1.1 Calm Water Resistance

RT [N ] is the total resistance, ⇢ [kgm 3

MARINTEK restistance decomposition

CT = CR + (1 + k) (CF + CF ) + CDB + CAA + CApp + CA (2.2)

where the dimensionless coefficients are:

CT = (1 + k)CF + CF + CA + CR + CAA (2.3)

• 10-25% for low-speed ships (such as bulk carriers and tankers)

• Newbuild vessel: 100 150µm

Based on Townsin and Byrne (1980):

• Newbuild vessel: 80 120µm

Shallow water correction

2.1.2 Added Resistance due to Wind and Waves

Rtot = RT + Rwind + Rwave (2.10)

Rtot - Total resistance in wind and waves

Added wave resistance

Rwave = RAWRL + RAWML (2.11)

Added resistance in wind

⇢A - Mass density of air

⌘0 - Open-water (propeller) efficiency

2.2.1 Propulsive Efficiency in Ideal Conditions

The hull efficiency ⌘H is defined by Equation 2.14

Relative rotative efficiency

2.2.2 Load Variation Correction

(RT + Rwave + Rwind ) · V

2.2.3 Total Delivered Power

⌘G - Gear box losses

2.3 Ship Powering Methods