Download as pdf or txt
Download as pdf or txt
You are on page 1of 120

-

Development and validation of hydrodynamic


model for near free surface maneuvers of BB2
Joubert generic submarine
Kim, Yagin
https://iro.uiowa.edu/esploro/outputs/9984097367102771/filesAndLinks?institution=01IOWA_INST&index=null

Kim, Y. (2021). Development and validation of hydrodynamic model for near free surface maneuvers of
BB2 Joubert generic submarine [University of Iowa]. https://doi.org/10.17077/etd.005772

https://iro.uiowa.edu
Free to read and download
Copyright 2021 Yagin Kim
Downloaded on 2023/03/15 21:55:11 -0500

-
Development and Validation of Hydrodynamic Model for Near Free Surface Maneuvers of BB2

Joubert Generic Submarine

by

Yagin Kim

A thesis submitted in partial fulfillment


of the requirements for the Doctor of Philosophy
degree in Mechanical Engineering in the
Graduate College of
The University of Iowa

May 2021

Thesis Committee: Pablo M. Carrica, Thesis Supervisor


Juan E. Martin, Thesis Supervisor
James H. J. Buchholz
Casey M. Harwood
Venanzio Cichella
Copyright by

Yagin Kim

2021

All Rights Reserved


ABSTRACT

The stability and maneuverability of underwater vehicles is still challenging today due to

their nature as neutrally buoyant, slender bodies mainly designed to move efficiently forward. The

maneuvering stability of a submarine requires properly designed control surfaces and controllers,

but even a submarine with well-designed controllers for normal operating conditions may

experience failure when operating below their design speed, or subject to disturbances from the

interaction with nearby interfaces (the free surface or the bottom of the sea). Therefore, developing

controllers capable of near-free surface operation is of great importance to prevent unintended

surfacing or diving that could expose the vehicle to damage or detection. Difficulties controlling

underwater vehicles increase as the speed decreases due to loss of control surface authority.

This thesis focuses on the development of a reduced order hydrodynamic model of a

submarine capable of simulation of near surface maneuvers, which is fast enough to be used for

development of controllers. While the study focuses on the generic submarine Joubert BB2, the

procedures are general and could be applied to other underwater craft, including submarines,

autonomous and unmanned underwater vehicles. The implemented model solves the equations of

rigid body motion subject to inertial forces and hydrodynamic loadings modeled as derivatives of

Taylor series expansions obtained from steady and unsteady computational fluid dynamics (CFD)

simulations and experimental data. The waves are implemented through a one-way coupling

approach where the pressure and virtual mass forces induced by the waves are computed from

unperturbed superposition of exact solutions for a given spectrum.

The hydrodynamic model of the Joubert BB2 submarine was derived from the original

Gertler and Hagen model (Gertler and Hagen 1967) for a cruciform stern plane arrangement. The

loading terms due to hull angle of attack and control surface performance correction terms due to

ii
in-plane rotations were modified to develop a more concise formulation. A set of numerical captive

model simulations were designed to obtain the acceleration-based added mass terms that are added

to the external loading terms separately. The free surface effects on derivatives were modeled by

parameterizing hydrodynamic coefficients with depth.

The model is implemented in the commercial software MATLAB SIMULINKTM to solve

the equations of motion implicitly. The hydrodynamic model was validated using model test results

from MARIN in Netherlands and CFD simulations, using the original Proportional-Integral-

Differential (PID) controller used in the CFD simulations and the experiments.

iii
PUBLIC ABSTRACT

Submarine maneuvering near surface experiences environmental loads induced by free

surface interaction and waves. During the maneuver in periscopic depth or confined water, the

vehicle is forced to move slow, leading to decreased control authority due to loss of hydrodynamic

efficiency of the control surfaces. Since the vehicle controllers are typically developed through

iterative process using reduced order models, a hydrodynamic model accounts for environmental

disturbance is essential to devise optimal control strategy.

In this study, a modified standard hydrodynamic model for a generic design BB2 is

formulated from a set of simple numerical captive simulations at several depths to account for the

influence of free surface. The control surface performance prediction is simplified from standard

formulation and the hydrodynamic coefficients for flow incidence angle is evaluated from single

coefficient set and the effective flow angle of attack. The virtual mass is measured from several

hull sections, allowing forces and moments estimated with only translational elements. The

hydrostatic load is computed from integrating hydrostatic pressure over coarse surface grid. Ocean

wave is computed based on significant wave height and frequency based on Brettschneider

spectrum. The wave load is implemented by virtual mass due to wave particle acceleration and the

hydrostatic pressure due to free surface elevation.

The model is implemented to a commercial software MATLAB SIMULINKTM to solve

rigid body motions implicitly. The hydrodynamic model is validated by validation study against

model experiment by MARIN in Netherlands, and CFD simulation results, using Proportional-

Integral-Differential controller with identical gains used by CFD simulations and experiments.

iv
TABLE OF CONTENTS

LIST OF FIGURES ..................................................................................................................... viii

LIST OF TABLES ......................................................................................................................... xi

NOMENCLATURE & ABBREVIATIONS ................................................................................ xii

1. INTRODUCTION ...............................................................................................................1

1.1. Motivation .....................................................................................................................1

1.2. Dynamic Modeling of Submarine Motions ...................................................................2

1.3. Free Surface and Wave loads ........................................................................................5

1.4. Controlled Submarine Maneuvering Simulations .........................................................7

1.5. Contributions of the Thesis............................................................................................9

2. GEOMETRY .....................................................................................................................12

2.1. Joubert BB2 .................................................................................................................12

2.2. Hull and Control Surfaces ...........................................................................................13

2.3. Propeller.......................................................................................................................14

2.4. Coordinate Systems .....................................................................................................14

3. CFD METHODS ...............................................................................................................17

3.1. CFD Code REX and Numerical Details ......................................................................17

3.1.1. REX ..................................................................................................................... 17

3.1.2. Momentum Equation and Turbulence Modeling ................................................ 17

3.1.3. Turbulence Modeling .......................................................................................... 18

3.1.4. Overset grids and motions ................................................................................... 20

3.1.5. Free Surface Modeling ........................................................................................ 21

3.1.6. Ocean Waves ....................................................................................................... 21

3.2. Propulsion and Controllers ..........................................................................................22

3.2.1. Propeller Cruise Control...................................................................................... 22

v
3.2.2. Vertical and Horizontal Controller ...................................................................... 23

3.2.3. Trim and Ballast Tank ......................................................................................... 24

3.3. Computational Grid and Boundary Conditions ...........................................................24

3.3.1. Grids .................................................................................................................... 24

3.3.2. Boundary Conditions........................................................................................... 26

4. DYNAMIC MODEL OF A SUBMARINE .......................................................................29

4.1. Rigid Body Motion ......................................................................................................29

4.2. Hydrodynamic Forces and Moments ...........................................................................30

4.2.1. Gertler and Hagen Model for X shape Stern Planes ........................................... 30

4.2.2. Model Reduction by Simplified High Order Correction ..................................... 37

4.2.3. Flow Angle of Attack Correction to Control Surfaces ........................................ 40

4.2.4. Control Surface Performance in Waves .............................................................. 42

4.2.5. Virtual Mass Forces and Moments...................................................................... 42

4.2.6. Virtual Mass Forces and Moments due to Wave Particle Acceleration .............. 43

4.3. Hydrostatic Forces and Moments ................................................................................44

4.3.1. Surface Grid ........................................................................................................ 44

4.3.2. Computation of Hydrostatic Forces .................................................................... 44

4.3.3. Computation of Hydrostatic Moments ................................................................ 45

4.3.4. Wave Hydrostatic Pressure ................................................................................. 46

4.4. Captive Simulations.....................................................................................................46

4.4.1. Self-Propulsion and Straight Towing Simulations .............................................. 47

4.4.2. Accelerated Forward Simulations ....................................................................... 51

4.4.3. Pure Drift Simulations......................................................................................... 53

4.4.4. Straight Towing with Static Control Surface Deflection .................................... 57

4.4.5. Pure Rotation Simulation .................................................................................... 67

vi
4.4.6. Rotating Arm Simulations................................................................................... 69

4.4.7. Propeller side forces and moments...................................................................... 73

5. VALIDATION OF HYDRODYNAMIC MODEL ...........................................................75

5.1. Free Roll Decay Test ...................................................................................................75

5.2. Vertical Zig Zag Maneuver (VZZ) ..............................................................................76

5.3. Horizontal Zig Zag Maneuver (HZZ) ..........................................................................80

5.4. Self-Propulsion in Waves ............................................................................................82

5.5. Horizontal Turning Circle Maneuver ..........................................................................83

5.6. Max-q Maneuver .........................................................................................................87

6. SUMMARY, CONCLUSIONS, LIMITATIONS AND FUTURE WORK ......................89

6.1. Summary......................................................................................................................89

6.2. Conclusions .................................................................................................................90

6.3. Limitations ...................................................................................................................91

6.4. Future Work .................................................................................................................92

REFERENCES ..............................................................................................................................93

APPENDIX. Hydrodynamic Coefficients Normalization .............................................................99

vii
LIST OF FIGURES

Figure 1. The geometry of Joubert BB2 ....................................................................................... 13

Figure 2. Propeller Open Water (POW) performance curve of MARIN 7371R .......................... 14

Figure 3. Local coordinate system and sign conventions for BB2 ............................................... 15

Figure 4. Computational domain and boundary conditions .......................................................... 25

Figure 5. The angle of attack correction for the upper portside stern plane ................................. 41

Figure 6. Computation of hydrostatic forces on the 𝑖th triangle ................................................... 44

Figure 7. Computation of hydrostatic moments for the 𝑖th triangle ............................................. 45

Figure 8. Thrust and resistance as function of approach speed and depth .................................... 48

Figure 9. Free surface elevation for self-propelled BB2 at 6kts and 10kts and two depths. ....... 49

Figure 10. Thrust deduction for approach speeds 3kts, 6kts and 10kts. ..................................... 50

Figure 11. Hull surface sections for virtual mass coefficient evaluation ...................................... 51

Figure 12. Hull resistance, vertical force and pitch moment as function of depth, speed and 𝛼 .. 54

Figure 13. Hull resistance and lateral and vertical forces and as function of depth, speed and
𝛽 .................................................................................................................................................... 55

Figure 14. Hull roll, pitch and yaw moments as function of depth, speed and 𝛽 ......................... 56

Figure 15. Sail plane resistance, vertical force, and pitch moment as a function of depth,
speed and deflection angle ............................................................................................................ 58

Figure 16. Lower starboard stern plane resistance, lateral force, and vertical force as a
function of depth, speed and deflection angle .............................................................................. 59

Figure 17. Lower starboard stern plane roll, pitch, and yaw moments as a function of depth,
speed and deflection angle ............................................................................................................ 60

Figure 18. Upper starboard stern plane resistance, lateral force, and vertical force as a
function of depth, speed and deflection angle .............................................................................. 61

Figure 19. Upper starboard stern plane roll, pitch, and yaw moments as a function of depth,
speed and deflection angle ............................................................................................................ 62

Figure 20. Upper portside stern plane resistance, lateral force, and vertical force as a function
of depth, speed and deflection angle ............................................................................................. 63

viii
Figure 21. Upper portside stern plane roll, pitch, and yaw moments as a function of depth,
speed and deflection angle ............................................................................................................ 64

Figure 22. Lower portside stern plane resistance, lateral force, and vertical force as a function
of depth, speed and deflection angle ............................................................................................. 65

Figure 23. Lower portside stern plane roll, pitch, and yaw moments as a function of depth,
speed and deflection angle ............................................................................................................ 66

Figure 24. Pure rotation simulations in vertical xy and yz planes ................................................ 67

Figure 25. Damping forces and moments due to q ....................................................................... 68

Figure 26. Damping forces (a) and damping moments (b) due to q ............................................. 68

Figure 27. Damping forces (a) and damping moments (b) due to 𝑢𝑝 .......................................... 69

Figure 28. Rotating arm simulations in 𝑦𝑧-plane (a) 𝑥𝑧-plane (b) 𝑥𝑦-plane (c) .......................... 69

Figure 29. Hydrodynamic forces (a) and moments (b) obtained from RA tests in 𝑥𝑧-plane ....... 72

Figure 30. Hydrodynamic forces (a) and moments (b) obtained from RA tests in 𝑥𝑦-plane ....... 72

Figure 31. Propeller side forces and moments (a) variation of thrust and resistance (b) as
function of speed ........................................................................................................................... 74

Figure 32. Propeller roll moment as function of speed and 𝑛 ...................................................... 74

Figure 33. Free roll decay test against model experiment ............................................................ 76

Figure 34. Time series of 𝜃 during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts. ......................... 78

Figure 35. Time series of 𝑞 during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts. ......................... 78

Figure 36. Time series of velocity during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts. .............. 79

Figure 37. Time series of depth change during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts. ...... 79

Figure 38. Time series of appendage total vertical force (𝑎) and total pitching moment (b)
from sail and stern planes at 10kts ............................................................................................... 80

Figure 39. The 20/20 HZZ motions in model test, CFD simulation and HDM simulation at
10kts ............................................................................................................................................. 81

Figure 40. First harmonic of depth change normalized by wave amplitude (a) and 𝜃 (b) from
self-propulsion in regular waves SS5 ........................................................................................... 82

Figure 41. Trajectory during a turning circle maneuver in calm water at 10kts .......................... 84

ix
Figure 42. Trajectory during turning circle maneuver in regular waves SS5 at 10kts ................ 85

Figure 43. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at the
sail top depth23.69 m as function of heading angle ..................................................................... 86

Figure 44. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at the
sail top depth 4 m as function of heading angle ........................................................................... 86

Figure 45. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at sail
top depth 2.5 m as function of heading angle ............................................................................... 87

Figure 46. The time series of motions during max-q maenuver: (a) 𝜃 (b) speed decrease (c)
pitch rate (d) vertical travel distance ............................................................................................. 88

x
LIST OF TABLES

Table 1. Main particulars of Joubert BB2 ..................................................................................... 16

Table 2. Nominal Sea State index and corresponding wave frequencies and heights .................. 22

Table 3. Details of computational grids ........................................................................................ 27

Table 4. Details of boundary conditions ....................................................................................... 28

Table 5. The list of surface grids used in hydrostatic forces and moments computation ............. 44

Table 6. Virtual mass coefficients obtained for different sections ............................................... 52

xi
NOMENCLATURE & ABBREVIATIONS

𝐴𝑖 , 𝑨𝒊 Area of a surface element and surface area normal vector, respectively


𝐴𝑤𝑎𝑣𝑒 Wave amplitude
𝐷0 Sail top depth of vehicle
𝐻𝑠 Significant wave height
𝐼𝑥𝑥 , 𝐼𝑥𝑥 , 𝐼𝑥𝑥 Mass moment of inertia about 𝑥, 𝑦, and 𝑧 axis respectively
𝐾 ′ , 𝑀′ , 𝑁 ′ Non-dimensional moments about 𝑥, 𝑦, and 𝑧 axis, respectively
𝐾𝑄 Propeller torque coefficient
𝐾𝑇 Propeller thrust coefficient
𝐾𝑛′ Roll moment coefficient due to propeller rotation
𝐿0 Characteristic length of vehicle
𝑇𝑝𝑟𝑜𝑝 Propeller thrust

𝑇𝑢(𝑞+𝑟) ′
, 𝑋𝑢(𝑞+𝑟) Hydrodynamic coefficients for thrust and resistance variation arise from
angular velocities
𝑈0 Approach speed of vehicle
𝑊0 Weight of vehicle
𝑋′, 𝑌′, 𝑍′ Non-dimensional linear forces in 𝑥, 𝑦, and 𝑧 axis, respectively
Lower starboard, upper starboard, upper portside, lower portside, stern
𝑋1, 𝑋2, 𝑋3, 𝑋4, 𝑋𝐵
planes and sail planes, respectively
′ ′ ′
𝑋𝑝𝑝 , 𝑌𝑝|𝑝| , 𝑍𝑝𝑝 ,
′ ′ ′ Hydrodynamic coefficients arise from angular velocity about 𝑥 axis
𝐾𝑝𝑝 , 𝑀𝑝𝑝 , 𝑁𝑝|𝑝|
′ ′ ′
𝑋𝑞|𝑞| , 𝑍𝑞|𝑞| , 𝑀𝑞|𝑞| Hydrodynamic coefficients arise from angular velocity about 𝑦 axis
′ ′ ′
𝑋𝑢|𝑝| , 𝑌𝑢𝑝 , 𝑍𝑢|𝑝| , Hydrodynamic coefficients arise from coupled freestream velocity and
′ ′ ′
𝐾𝑢𝑝 , 𝑀𝑢|𝑝|, 𝑁𝑢𝑝 angular velocity about 𝑥 axis
′ ′ ′
𝑋𝑢|𝑟| , 𝑌𝑢𝑟 , 𝑍𝑢|𝑟| , Hydrodynamic coefficients arise from coupled surge velocity and
′ ′ ′
𝐾𝑢𝑟 , 𝑀𝑢|𝑟| , 𝑁𝑢𝑟 angular velocity about 𝑧 axis in 𝑥𝑦 plane

′ ′ ′ Hydrodynamic coefficients arise from coupled surge velocity and


𝑋𝑢𝑞 , 𝑍𝑢𝑞 , 𝑀𝑢𝑞
angular velocity about 𝑦 axis in 𝑥𝑧 plane
′ ′ ′
𝑋𝑢𝑣𝑤 , 𝑌𝑢𝑣𝑤 , 𝑍𝑢𝑣𝑤 ,

𝐾𝑢𝑣𝑤 , Hydrodynamic coefficients arise from hull drift
′ ′
𝑀𝑢𝑣𝑤 , 𝑁𝑢𝑣𝑤

xii
′ ′ ′ Hydrodynamic coefficients arise from coupled heave velocity and
𝑋𝑤𝑞 , 𝑍𝑤𝑞 , 𝑀𝑤𝑞
angular velocity about 𝑦 axis in 𝑥𝑧 plane
′ ′ ′
𝑋𝛿𝑋 𝑖
, 𝑌𝛿𝑋 𝑖
, 𝑍𝛿𝑋 𝑖
,
′ ′ ′ Appendage coefficients arise from deflection angle of 𝑖th control surface
𝐾𝛿𝑋𝑖 , 𝑀𝛿𝑋𝑖 ,𝑁𝛿𝑋𝑖

′ ′ ′ ′ Hydrodynamic coefficients for propeller side forces and moments arise


𝑌𝑢𝑟 , 𝑍𝑢𝑞 , 𝑀𝑢𝑞 , 𝑁𝑢𝑟
from angular velocities
𝑐𝑣𝑚,𝑥𝑥 , 𝑐𝑣𝑚,𝑦𝑦 ,
Virtual mass coefficients in 𝑥, 𝑦 and 𝑧 axis, respectively
𝑐𝑣𝑚,𝑧𝑧
𝑘𝑤𝑎𝑣𝑒 Wave number
Pressure measured at the center of 𝑖th and 𝑗th node of a triangular
𝑝𝑖𝑗
surface element
𝑡𝑑 Propeller thrust deduction
𝑭𝑖 Force vector of 𝑖th triangle
𝑭𝑣𝑚 Virtual mass forces
𝑴𝑣𝑚 Virtual mass moments
𝑼𝑟,𝑖 Rigid body velocity of 𝑖th body element in ship coordinate system
𝒏𝑖 Unit normal vector of 𝑖th surface element
𝒓𝑋𝑖 Position vector of geometric center of 𝑖the control surface from 𝐶𝐺
Position vector to the center of 𝑗th and 𝑘th node of 𝑖th surface element
𝒓𝑗𝑘,𝑖
from the 𝐶𝐺
𝒖𝑝,𝑋𝑖 Projection of 𝒖𝑠,𝑋𝑖 to the plane of 𝑖th control surface at neutral deflection
Velocity vector defined at the geometric center of 𝑖th control surface in
𝒖𝑠,𝑋𝑖
ship coordinate system
𝒖𝑠 Velocity vector defined at 𝐶𝐺 in ship coordinate system
Wave particle velocity vector defined at the 𝐶𝐺 in ship coordinate
𝒖𝑤𝑠
system
𝛀𝑠 Angular velocity vector in ship coordinate system
𝛼𝑋𝑖 Flow angle of attack correction to 𝑖th control surface
𝛿𝑉 , 𝛿𝐻 Vertical and horizontal actuator angle, °
𝜂0 Propeller efficiency
𝜔𝑠 Significant wave frequency, 𝑟𝑎𝑑/𝑠
𝜔𝑤𝑎𝑣𝑒 Wave frequency

xiii
ROM Reduced order model
𝐷 Diameter
𝐷0 Depth of sail
𝐻𝐷𝑀 Hydrodynamic model
𝐽 Propeller advance coefficient
𝐾, 𝑀, 𝑁 Moments about 𝑥, 𝑦, and 𝑧 axis, respectively
𝐿𝐶𝐺 Longitudinal center of gravity
𝑅 Turning radius
𝑆(𝜔) Energy density of spectrum, 𝑚2 𝑠
𝑆𝑆 Sea state
𝑋, 𝑌, 𝑍 Linear forces in 𝑥, 𝑦, and 𝑧 axis, respectively
𝑔 Acceleration due to gravity
𝑛 Propeller rotation rate, 𝑟𝑒𝑣/𝑠
𝑝 Pressure
𝑡 Propeller thrust deduction
𝑴 Mass matrix
𝒔 Velocity state vector
𝛥𝑚𝑖 Fluid mass equivalent to displaced volume of 𝑖th body component
𝜌 Density of fluid

xiv
1. INTRODUCTION

1.1. Motivation

Modeling of submarine maneuvers under environmental disturbances such as ocean

currents and waves can be a very challenging problem (Fang et al. 2006, Teo et al. 2012). When

at periscope depth, the propulsion speed may be slowed down to the reverse speed (or dive plane

reversal point, or the speed point where the vertical force when actuating the stern planes changes

sign, Papoulias and Riedel 1994), where the vehicle controllability is compromised by reduced

control plane authority and inability to use the hull pitch to control depth (Mohseni 2006). Littoral

operations or maneuvers in restricted waters also force the vehicle to move slow (Broglia et al

2006, Nematollahi et al. 2015). Development of controllers and autopilots for underwater vehicles

is typically done using reduced order models (ROM) for the dynamics of the craft which can

provide a testbed to run simulations very fast at the cost of reduced accuracy. Free running

experiments (Overpelt et al. 2015) or sophisticated computational fluid dynamics (CFD)

simulations (Carrica et al. 2020) may also predict motions incorporating controllers and autopilots,

but they are considerably more expensive and time consuming. Early development of control

strategies and algorithms and subsequent initial tuning are typically performed using ROM,

leaving experiments and CFD for fine tuning or to study controller performance under more

complex effects not captured by simpler ROM approaches.

In this thesis a hydrodynamic ROM is developed and validated for the generic Joubert BB2

submarine. The model coefficients are obtained from a set of CFD simulations with imposed linear

and angular velocities and accelerations with the submarine at different depths. Free surface and

wave forces and moments are computed in a one-way coupling approach, in which waves influence

the motions of the submarine but the waves are not affected by its presence. While previous

1
proposed submarine dynamic models have no or limited consideration of free surface presence and

wave effects, the model presented in this thesis incorporates these effects by computing the

coefficients as a function of depth.

1.2. Dynamic Modeling of Submarine Motions

Typical dynamic models of marine vehicles are composed of equations describing the rigid

body motions, estimation of the hydrodynamic loads, and implementation of external disturbances

like currents and waves. The computation of the hydrodynamic loads is critical as the behavior of

the model depends heavily on them. Hydrodynamic force coefficients (also called derivatives) are

typically expressed as function of the body motions state by a Taylor series expansion, but to

exactly describe the flow around the object an unmanageable number of terms would be required.

Consequently, only the key terms are maintained and found by an optimization process like the

least square methods (Go and Ahn 2019), though how many and which terms to keep is decided

based on relative importance for the maneuver or application and the complexity to compute its

derivative. Gertler and Hagen (1967) proposed a set of equations for generic submarines with

cruciform stern planes, including the main terms influencing generic maneuvers. Feldman (1987)

modified the equations to include effects of crossflow, however the original form has been

preferred by the research community.

From the steady state assumption, the hydrodynamic forces and moments acting on the

body are only dependent on the magnitude of the corresponding motion. Each term in the model

is then the contribution of the corresponding motion. This assumption allows the forces and

moments to be calculated without the history of motions, and only requires time-independent

velocities and accelerations. This is convenient to develop the model, but it is also an inherent

factor that limits the accuracy of prediction of extreme maneuvers where unsteadiness is

2
significant. The hydrodynamic derivatives, the coefficients describing hull loading and appendage

forces and moments, are often assumed to be linear when the body is hydrodynamically stable or

expected to have only small incidence flow angles throughout the mission (Evans and Nahon 2004,

Phillips et al. 2010). The derivatives are evaluated by using a test matrix covering normal operating

conditions (Feldman 1995). There are three major approaches in evaluation of model coefficients:

experimental, analytical and semi-empirical (ASE), and computational methods.

Experimental evaluation of coefficients has been conducted using captive model tests and

free-running experiments. For captive model tests, a scaled model is attached to a vertical or

horizontal Planar Motion Mechanism (PMM) or a rotating arm (RA) to obtain the coefficients.

Experimental techniques and analysis methods are found from early research (Goodman 1960).

Periodic sinusoidal and linear motions are imposed to vertical and horizontal PMM tests to

evaluate transient (acceleration-based) coefficients. Steady turning circle motions with or without

angles of attack are imposed in RA test. The set of imposed motions is chosen to isolate coefficients

from one another. Early research with measurements and stability analysis of control

characteristics from straight line towing and RA can be found in Feldman (1987, 1995). Watt and

Bohlman (2004) studied the roll stability of a rising submarine from results of a captive model test.

Static tests such as straight line towing or with hull incidence angle can also be replaced by wind

tunnel tests; however the achievable Reynolds number in water and air is often different (Park et

al. 2017). Free-running tests of submarines use remotely operated scale models to conduct

maneuvering simulations, typically to investigate overshoots in vertical and horizontal plane

maneuvers and turning circle maneuvers in the horizontal plane. The controller of the submarine

should be ready before the free running test; therefore, the purpose of this simulation is typically

not to obtain hydrodynamic coefficients. Free-running model tests (e.g. Overpelt et al. 2015)

3
provide good benchmark data to unsteady controlled CFD maneuvering simulations (Carrica et al.

2016 and 2019, Kim et al. 2018, Kim et al. 2020). The drawbacks of the experimental method in

addition to its cost and time is that the propeller interaction with the hull and rear control surfaces

is difficult to measure.

Analytical approaches use strip theory (Kim et al. 1980) to estimate the hydrodynamic

coefficients of a body. The object is dissected into continuous 2D cross sections, and the

hydrodynamic loads analytically approximated from each section are integrated along the body.

Since strip theory assumes that the object is slender, the vessel needs to have a small beam-length

ratio (Isa et al. 2014). In semi-empirical methods, the geometry is separated into smaller

components and analyzed as simpler shapes, using experimentally derived or empirical

relationships based on parameters like aspect ratio to obtain the hydrodynamic coefficients of each

component. The method yields reasonable estimation for generic shapes and conventional designs,

but it has limitations against novel configurations (Geisbert 2007).

Computational methods can be categorized into potential and CFD methods. Potential

theory calculates the coefficients based on velocity potentials and provides a cost-effective means

of model evaluation. Since the acceleration terms cannot be obtained from steady-state CFD

simulations, potential theory is sometimes chosen as alternative method for those coefficients

(Watt 2007), but it is not capable of estimating dissipative terms. Due to this theoretical limitation,

the potential theory is reported to be inaccurate for large incidence angles (Evans et al. 2004).

Steady-state CFD simulations have been used as an alternative to experimental methods in recent

years. It has been mainly applied to research of unmanned underwater probes with a simple

geometry. Early applications replaced towing test to steady state CFD simulations with fixed grids

4
(Tyagi and Sen 2006). Experiments and CFD calculations are in some research used together

(Wang et al. 2014).

Numerical towing and RA tests were used to compute coefficients in the early literature.

Since the unsteady nature of PMM test requires more computational power and capabilities, early

efforts did not use numerical PMM to evaluate hydrodynamic forces. With advances in

computational techniques and power, virtual PMM simulations using RANS have also been

conducted (Zhang et al. 2010, Pan 2012, Sakamoto et al. 2012). Gao et al. (2018) performed

numerical PMM tests to identify the hydrodynamic coefficients of the DARPA Suboff and a small

size UUV, simplifying the model equations using the R-square method. The coefficients from CFD

results were compared against the experimental result with good agreement. Cho et al. (2020)

constructed a hydrodynamic model based on PMM CFD simulations and performed a turning

circle maneuver at depth. One notable advantage of CFD over experiments is that it can be used

to evaluate hull/propeller and stern planes/propeller interactions (Pan et al. 2019). The application

of CFD for coefficient evaluation, however, has been limited to an alternative to PMM or RA test,

and a self-consistent set of tests to evaluate the coefficients for submarine dynamic model

equations has not been developed.

1.3. Free Surface and Wave loads

The influence of free surface and waves on hydrodynamic loads on underwater vehicles

has been studied by researchers from the naval architecture, ocean engineering/science, and control

communities.

Numerical studies on free surface effects on hydrodynamic forces of submerged bodies

have mainly focused on calm water conditions. Mansoorzadeh and Javanmard (2014) carried out

a numerical and experimental study of the free surface effect drag and lift coefficients of a towed

5
AUV. Dawson (2014) studied experimentally and numerically the control authority of a

conventional submarine operating near the surface, reporting that a rapid decrease with the

distance from the free surface. Gourlay and Dawson (2015) investigated the capability of potential

flow methods to predict wave making resistance of DARPA Suboff and Joubert near the free

surface. The results shows that the method yields good wave resistance predictions for Suboff and

slight overprediction for Joubert at relatively high Froude numbers and small body slenderness

ratio in calm water. Shariati and Mousavizadegan (2017) studied the influence of the free surface

to the performance of the appendages of DARPA Suboff by investigating viscous, pressure and

wave making resistance using the commercial code ANSYS CFX. The study concludes that the

presence of appendages does not have significantly affect wave resistance. Mansoorzadeh and

Javanmard (2014) investigated free surface effects as a function of distance to the free surface for

an AUV, but clear trends were not established.

More recent CFD simulations offer more complex conditions. Tian et al. (2019) studied

the influence of surface waves on a bare UAV in regular waves. Carrica et al. (2016, 2018)

simulated free running and constrained CFD simulations to investigate surface wave loads on the

vehicle and impact on performance of a PID controller for Joubert BB2. The studies have

confirmed that the influence of the free surface and waves decrease as the vehicle operates deeper,

and that virtual mass forces induced by waves are significant.

Analytical models of surface wave disturbances are rarely found from the literature of the

control community. Fang et al. (2006) investigated the influence of wave forces on the

maneuverability of NPS AUV II during ascending and descending maneuvers. The surface wave

forces and moments were modeled using strip theory to integrate incident wave and diffraction

potentials. ROM simulations were performed showing that the additional disturbances due to

6
surface waves deteriorate the performance of the PD controller as achieves a target depth, though

no experimental of CFD validation is provided. Another example of strip theory to incorporate the

wave force into a maneuvering model is given by Wang et al. (2015), where a AUV roll motion

model including wave loads was used to design a controller to minimize the roll motion. These

studies have in general have implemented the wave drag force as function of the first derivative of

the wave potential. In this approach a substantial disturbance is provided to change the behavior

of controllers, but wave forces due to virtual mass and wave induced pressures are not properly

accounted. Fischer et al. (2014) is an example of modelling wave loads as first and second time

derivatives, but detailed formulation is not provided.

1.4. Controlled Submarine Maneuvering Simulations

Unlike common underwater Unmanned Autonomous Vehicles (UAV) with mostly

axisymmetric body shapes, conventional submarines have hydrodynamically unstable shapes. The

hull is not symmetric about the horizontal plane and the coning tower or the sail structure

consistently generates bow-up pitching moment. This unstable nature requires more frequent

control surface actuation for course keeping and other attitude compensating mechanisms such as

trim and ballast tanks.

Advances in computing capabilities have made large scale unsteady CFD simulations of

maneuvers possible. Chase et al. (2013) was the first to perform simulations with controllable stern

planes and rotating propeller of the DARPA Suboff performing turning and surfacing maneuvers.

Simulation results of vertical and horizontal maneuvers for the Joubert BB2 submarine using

discretized propeller grid (Carrica et al. 2020) were in good agreement with free-running model

experiments of Overpelt et al. (2015). Near free surface self-propulsion and turning maneuvers in

regular and irregular wave was studied further in Carrica et al. (2019) and Kim et al. (2020).

7
Benchmark simulations of experimental results by Overpelt et al. (2015) was also conducted by

Kim et al. (2018) using a body force propeller model. Martin et al. (2015) performed vertical and

horizontal overshoot maneuvers of the ONR Body-1 submarine using body force and discretized

rotating propeller approaches, showing that the propeller model did not affect the results of

maneuvers significantly. Other significant efforts to simulate direct maneuvers of Joubert BB2

include near surface self-propulsion (Carrica et al. 2016), and vertical zigzag maneuvers (Carrica

et al. 2021).

The controller performance becomes more important when additional external

environmental forces exist. In the presence of a boundary or interface near the hull, suction forces

and moments may occur. A submarine or underwater vehicle operating near the seabed or any

underwater terrain is always exposed to the risk of collision due to such effect (Crook 1994).

Likewise, a submarine in littoral operation or approaching harbor will experience highly nonlinear

interactions with the free surface. Additionally, near the free surface, a low pressure region in the

propeller wake generates a bow down pitching moment and that makes maintaining target depth

more difficult. Under these conditions the depth controller will try to lower the vehicle by actuating

bow and stern planes, however these control surfaces might not be sufficient to achieve both the

target depth and attitude (Carrica et al. 2016). Operation of trim and ballast tank controllers can

provide a solution to maintaining the target depth and attitude at the same time. The ballast tank

increases or decreases the displacement of the submarine by pumping in or out water from the

system, and the trim tank controller moves the longitudinal center of gravity by moving water

between tanks located near the bow and stern. There are practical limitations to use of trim and

ballast tanks, including limits in maximum flow rate of the pumps and cost of operation of ballast

tank.

8
Operation near the surface adds another challenge to controllability. Wave making

resistance occurs when the submarine sails near the free surface. Operation in waves exposes the

submarine to a random flow fields, requiring continuous actuation of controllers. Pitch and heave

control are significantly affected, and the pitch angle in particular may adversely affect the average

performance of the propeller by disrupting the propeller wake (Carrica et al. 2019); a similar

phenomenon was explained by Faltinsen et al. (1980) for surface ships.

For “X” shaped configuration of the stern planes, depth and pitch control cannot be

achieved without affecting the horizontal maneuvering performance. The sway force and yawing

moment are generated from all four stern planes, and each of them contribute to both horizontal

and vertical maneuvering performance at any deflection angle (Overpelt et al. 2015, Carrica et al.

2016), therefore a change in vertical command brings in at the same time a change in sway force

and yawing moment. Due to frequent actuation of the vertical controller at periscope depth

operation in waves, changes in horizontal maneuvering performance are also frequent. As

observed in Kim et al. (2020), the heading angle in a seaway has great influence in seakeeping

ability and propulsion; similarly, the tactical diameter of a steady turning maneuver with fixed 𝛿𝐻

is also affected by the wave encounter angle.

1.5. Contributions of the Thesis

This thesis resulted in three main contributions. First, it produced a validated, free

distribution ROM of the generic submarine Joubert BB2 that incorporates free surface and waves

effects. Second, it established a set of tests to effectively isolate hydrodynamic coefficients for

ROMs. Third, it proposed a methodology to incorporate ocean wave loads into submarine ROMs.

In the first main contribution, the ROM of Joubert BB2 was implemented in the

commercial program Matlab SIMULINKTM. The model was used transitioned to Professor

9
Venanzio Cichella and his research group at The University of Iowa for developing adaptive

controllers for submarine maneuvering at different depths operating with speeds close and above

the reversal speed. The controller was developed as a result of the project supporting this thesis

and resulted in a published paper (Rober et al. 2021a) and another two in preparation (Kim et al.

2021 and Rober et al. 2021b). The hydrodynamic coefficients that comprise the ROM were

evaluated from single-phase steady state and unsteady CFD simulations using the unsteady

RANS/DES code REX developed at The University of Iowa. The coefficients are evaluated from

specifically designed captive test simulations including straight towing, even keel self-propulsion,

pure rotation, rotating arm and PMM tests as discussed in §3.4. The influence of the free surface

on the loading terms is modeled by parametrizing the coefficients as functions of depth. For CFD,

the free surface is modeled using a level set approach, as briefly described in §3.1. By performing

numerical captive simulations with various depths, the interaction with the free surface is

implemented through linear interpolations between computed depths.

In the second contribution of the thesis, a set of captive simulation cases are introduced to

decouple hydrodynamic loadings. Towing simulations with or without acceleration are used to

evaluate hydrodynamic coefficients for hull drift, control surface performance, propeller thrust

deduction and virtual mass. The virtual mass coefficients are obtained from several sections of the

vessel, enabling the estimation of virtual mass forces in waves by the ROM. RA tests are used to

evaluate the hydrodynamic coefficients arising from turning motions. Pure rotation tests are

introduced to isolate forces and moments due to rotation from results of RA test. The case set does

not require PMM tests, thus the estimation of hydrodynamic coefficients is done more directly and

independently.

10
For the third contribution, model of propagating regular waves was implemented. The

wave load includes the virtual mass forces and moments due to wave particle acceleration, and the

wave-induced pressure loads. The evaluated wave acceleration corrects the sectional body

acceleration to obtain reasonable total virtual mass load in a time-efficient manner. The wave

pressure loads are computed by integrating the wave-induced pressure forces over a coarse surface

mesh as described in §3.6, to evaluate the net buoyancy change due to free surface elevation.

The use of full scale CFD simulation data benefits the modeling process by eliminating

scale effects compensation. The flow around the propeller and the stern planes in the model test is

influenced by the scale effect due to large Reynolds number difference (Carrica et al. 2021),

therefore requires careful extrapolation. The propeller forces and moments reported from full-scale

free running vertical and horizontal zig zag CFD simulations are used to model the corresponding

coefficients. Using full-scale simulation results simplifies the overall derivation of the ROM.

11
2. GEOMETRY

2.1. Joubert BB2

Joubert BB2 is a publicly available research purpose generic SSK class submarine

geometry, the latest update to Joubert BB1. The Joubert geometry was first introduced by Joubert

(2004, 2006) as a result of a research project from the Australian Defense Science and Technology

Office (DSTO) for a conceptual design to their next generation submarines which will be needed

by 2026. Joubert designed the hull of Submarine 2026 based on existing geometries, assuming the

mission requirement will remain similar to those of the Collins class. Joubert focused on reducing

the resistance of the craft while retaining the volume for machineries required by the Collins class.

His design decisions are often based on Gertler (1950), which has been the technical foundation

of US Navy’s research submarine USS Albacore. With the arbitrarily set length of 70.0m and beam

length of 9.6m, the hull shape became shorter and wider than that of a Collins class craft. Joubert’s

ideal 𝐿/𝐷 was between 6 and 7, however it could not be achieved for the required machinery

volume and ended up at 7.3. The nose shape was obtained from a NACA foil section with thickness

14.3% of the length, to minimize the interference to the forward passive sonar array. The stern

planes employed an X shape arrangement similar to the Collins class. Joubert left the location and

shape of the front control surface to the future study. The first design did not have a clear cross

section shape of the sail either. Later a sail cross section of NACA 0017 was added with sail plane

installed in front, designating the submarine 2026 as BB1. This BB1 was revealed to be

directionally unstable, therefore the area of rear control surfaces was increased, and the sail and

sail planes were shifted forward (Overpelt et al. 2015). In addition, it was viewed not realistic to

use a NACA 0017 foil for the sail, therefore the thickness was increased to a NACA 0022.

12
BB2 aimed to achieve optimized sailing performance. With use of the computer program

SAMSON (Safety And Manoeuvring Simulation Of Naval submarines), the location of sail and

sail plane, and stern plane span were optimized by Overpelt et al. (2015). To improve horizontal

and vertical stability, and control surface efficiency, sail and sail planes were shifted forward and

the span of stern planes was increased. The reversal speed, the minimum speed where the

submarine can be controlled by the control surface is decreased due to the modifications. The BB2

was shown to have sub-recommended vertical stability, but acceptable horizontal stability and

control surface efficiency.

2.2. Hull and Control Surfaces

Figure 1 shows the geometry of Joubert BB2 with some key dimensions in full scale, and

details of main particulars are summarized in table 1 for full and model scale for the experiment

dataset from Overpelt et al. (2015).

Figure 1. The geometry of Joubert BB2

13
2.3. Propeller

The MARIN 7371R propeller was used for this study. The propeller hub and boss cap are

included. The diameter is 5.0𝑚 in full scale and 0.2725𝑚 in model scale. The open water curves

for MARIN 7371R are shown in Figure 2.

Figure 2. Propeller Open Water (POW) performance curve of MARIN 7371R

2.4. Coordinate Systems

The local coordinate system for the vehicle is shown in the Figure 2. The coordinate system

used in the dynamic model is identical to the CFD reference system, where the x-axis is directed

to the stern and the z-axis upward. Therefore, positive ship resistance is generated when the ship

moves forward, while thrust has a negative sign. Likewise, bow up pitch (𝜃 ), upward heave, roll

(𝜙) and yaw (𝜓 ) to the portside have all positive signs. The origin of the local coordinate is

located at the center of gravity 𝐶𝐺.

14
Figure 3. Local coordinate system and sign conventions for BB2

The velocities 𝑢, 𝑣, 𝑤, 𝑝, 𝑞 and 𝑟 and accelerations 𝑢̇ , 𝑣̇ , 𝑤̇ , 𝑝̇ , 𝑞̇ and 𝑟̇ are represented in

the local coordinate system and displacements 𝑥, 𝑦, 𝑧, 𝜙, 𝜃 and 𝜓 are represented in the global or

earth coordinate system.

The axis of propeller rotation is directed toward to the bow of the vessel. When viewed

from behind the submarine, positive rate of rotation is in the clockwise direction to generate thrust

that pushes the vehicle forward.

The axes of rotation of the six control surfaces a pair of sail planes and four stern planes

are illustrated in figure 3. The axis of the sail planes is directed to port, where positive 𝛿𝑋𝐵 means

upward lift sail plane deflection. The axes of the stern planes are directed outward. The stern plane

axes are slightly inclined to the aft., rather than perpendicular to the shaft. Since the stern planes

are labeled increasing counterclockwise from the lower starboard plane when viewed from the

behind, positive 𝛿𝑋1 and 𝛿𝑋2 means upward lift deflection and 𝛿𝑋3 and 𝛿𝑋4 have opposite sign

15
for deflection to the same direction. The actuation commands for control surfaces are implemented

as (Overpelt et al. 2015)

𝛿𝑋𝐵 = −𝛿𝑉
𝛿𝑋1 = −𝛿𝑉 − 𝛿𝐻
𝛿𝑋2 = −𝛿𝑉 + 𝛿𝐻 (1)

𝛿𝑋3 = 𝛿𝑉 + 𝛿𝐻
𝛿𝑋4 = 𝛿𝑉 + 𝛿𝐻

where 𝛿𝑉 and 𝛿𝐻 are the vertical and horizontal commands imposed by the controller discussed

in §3.2.

Table 1. Main particulars of Joubert BB2

Description Full Scale MARIN

Main Particulars

𝜆𝑠 Scale 1 18.348

𝐿0 Length overall (𝑚) 70.2 39.8631

𝐵0 Breadth of the hull (𝑚) 9.6 0.5232

𝑊0 Displacement (𝑡𝑜𝑛𝑠) 4453.6 0.7210

𝐶𝑄 Fullness parameter 0.85

𝐼𝑥 x-directional Moment of Inertia (𝑘𝑔 ∙ 𝑚2 ) 5.2481 ∙ 107 25.2382

𝐼𝑦 y-directional Moment of Inertia (𝑘𝑔 ∙ 𝑚2 ) 1.3794 ∙ 109 663.3559

𝐼𝑧 z-directional Moment of Inertia (𝑘𝑔 ∙ 𝑚2 ) 1.3673 ∙ 109 657.5369

𝐿𝐶𝐺 Longitudinal Center of Gravity form the Nose 32.31 1.7610

Propeller

𝑁𝑃 Number of blades 5

𝐷𝑃 Propeller diameter (𝑚) 5.0 0.2725

16
3. CFD METHODS

3.1. CFD Code REX and Numerical Details

3.1.1. REX

REX is a Reynolds-Averaged Navier-Stokes (RANS), Large Eddy Simulation (LES) and

hybrid RANS/LES CFD solver, under continuous development by Pablo Carrica’s Research

Group (PCRG) at the University of Iowa. REX is intended for high-Reynolds number flows,

causing the transport and reinitialization equations to be weakly elliptical and thus enabling penta-

diagonal line solvers in an alternate-direction-implicit (ADI) scheme to be used. Parallel

processing is enabled by use of an MPI-based domain decomposition, where each decomposed

block is mapped to one processor. Inter-processor information transfer is made after each ADI

iteration and boundary condition enforcement. The matrix for pressure includes the information of

the multi-block overset interpolation and boundary conditions. The pressure Poisson equation is

solved using a multigrid preconditioned Krylov solver from the PETSc library (Carrica et al.

2010). More advanced multi-body dynamic simulations are possible including cable connections

between bodies and also by linking with external multi-body dynamics solvers (Behara et al. 2020,

Li et al. 2017). Fluid-structure interaction problems can be simulated for flexible bodies using a

modal analysis outputted by a code like ANSYS or Abaqus (Paik et al. 2009). One-way or two-

way coupling is possible. Modeling of nonlinear deformations are done with an internal solver or

by coupling REX to a multibody solver (Li et al. 2015).

3.1.2. Momentum Equation and Turbulence Modeling

Continuity and momentum conservation are enforced solving the Navier-Stokes equations

for an incompressible fluid

17
𝜕𝑢𝑗
=0
𝜕𝑥𝑗
(2)
𝜕𝑢𝑗 𝜕𝑢𝑗 𝜕𝑝 𝜕 1 𝜕𝑢𝑖 𝜕𝑢𝑗
+ 𝑢𝑖 =− + [ ( + )] + 𝑠𝑖
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑖 𝜕𝑥𝑗 𝑅𝑒𝑒𝑓𝑓 𝜕𝑥𝑗 𝜕𝑥𝑖

where 𝑝 = 𝑝𝑎𝑏𝑠 ⁄𝜌𝑈02 + 2𝑘⁄3 is the dimensionless piezometric pressure, 𝑝𝑎𝑏𝑠 is the absolute

pressure𝑘 is the turbulent kinetic energy (TKE), 1⁄𝑅𝑒𝑓𝑓 = 1⁄𝑅𝑒 + 𝜈𝑡 is the effective Reynolds

number and Re is the Reynolds number based on the characteristic length and speed, 𝜈𝑡 is the

turbulent viscosity determined by the turbulence model and 𝑠𝑖 is the source term. The equations

are non-dimensionalized using a reference velocity 𝑈0 (chosen as ship velocity, experimental

tunnel velocity, etc.) and characteristic length L (ship/submarine length, propeller radius, etc.) .

Second order backward scheme is typically used for temporal discretization (other first

order schemes are also available). The convection terms in the momentum equation are discretized

with up to fourth order upwind biased scheme. Continuity is enforced, and pressure and velocity

fields are coupled using a projection method.

3.1.3. Turbulence Modeling

Turbulence is modeled using Menter’s blended k-ω/k-ε (Menter, 1994), or other models

(Spalart-Allmaras or Wilcox). The model equations are

𝜕𝑘 𝜕𝜈𝑡 𝜕𝑘 1 𝜕 2𝑘
+ (𝑢𝑗 − 𝜎𝑘 ) − + 𝑠𝑘 = 0
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝑃𝑘 𝜕𝑥𝑗2
(3)
2
𝜕𝜔 𝜕𝜈𝑡 𝜕𝜔 1 𝜕 𝜔
+ (𝑢𝑗 − 𝜎𝜔 ) − + 𝑠𝜔 = 0
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝑃𝑘 𝜕𝑥𝑗2

18
where 𝑘 is the turbulent kinetic energy (TKE) and 𝜔 the specific dissipation rate, with the
1
turbulent viscosity defined as 𝜈𝑡 = 𝑘/𝜔. The Peclet numbers are defined as 𝑃𝑘 = 1 , 𝑃𝜔 =
+𝜎𝑘 𝜈𝑡
𝑅𝑒

1
1 . The source for 𝑘 and 𝜔 are defined as following.
+𝜎𝜔 𝜈𝑡
𝑅𝑒

𝑠𝑘 = −𝐺 + 𝛽 ∗ 𝜔𝑘

𝐺 𝜕𝑘 𝜕𝜔 1 (4)
𝑠𝜔 = 𝜔 (𝛽 ∗ 𝜔 − 𝛾 ) − 2(1 − 𝐹1 )𝜎𝜔2 ( ) ( )
𝑘 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜔

Where the intermediate variables 𝐺, 𝐹1 and𝐶𝐷𝑘𝜔 are defined as

𝜕𝑢𝑖
𝐺 = 𝜈𝑡 𝜏 ∶ ( )
𝜕𝑥𝑗

4
√𝑘 1 500 4𝜎𝜔2 𝑘
𝐹1 = tanh [(min (max ( , ), )) ] (5)
0.09𝜔𝛿 𝑅𝑒 𝛿 𝜔 𝐶𝐷𝑘𝜔 𝛿 2
2

1 𝜕𝑘 𝜕𝜔
𝐶𝐷𝑘𝜔 𝛿 2 = max (2𝜎𝜔2 ( ) ( ) , 10−20 )
𝜔 𝜕𝑥𝑗 𝜕𝑥𝑗

For turbulence convection, first-order upwind scheme is used with linear single order

discretization. For high Re problems, a 3-layer wall function model is used for the boundary layer.

In the wake region, the blending function 𝐹1 is zero and switches to one in the logarithmic

and sublayer regions of the boundary layer. The distance to the closest no-slip surface, 𝛿, is needed

to calculate 𝐹1 . An SST model is an available option that accounts for turbulent stress transport.

This is useful to improve results for flows with adverse pressure gradients. The model differs from

Menter’s blended model by using the absolute value of the vorticity, 𝛺, to define the turbulent

viscosity like following.

0.31𝑘
𝜈𝑡 = (6)
max(0.31𝜔, Ω𝐹2 )

19
2
2√𝑘 500𝜈
𝐹2 = tanh [(max ( , )) ]
0.09𝜔𝛿 𝛿 2 𝜔

For massively separated flows, turbulence can be modeled using hybrid RANS/LES

models like the k-ω/k-ε based detached eddy simulation (DES) and delayed DES (DDES), or

directly Smagorinsky or Dynamic Smagorinsky LES. The dissipative term of the 𝑘 transport

equation is replaced by

𝑘 ∗
𝜌𝑘1.5
𝐷𝑅𝐴𝑁𝑆 = 𝜌𝛽 𝑘𝜔 =
𝑙𝑘−𝜔
(7)
1.5
𝑘
𝜌𝑘
𝐷𝐷𝐸𝑆 =
𝑙
𝑘 1.5
Where the length scales are defined as 𝑙̃ = min(𝑙𝑘−𝜔 , 𝐶𝐷𝐸𝑆 Δ) and 𝑙𝑘−𝜔 = 𝛽∗𝜔

respectively. 𝐶𝐷𝐸𝑆 = 0.65 and Δ is the local grid spacing. This model determines where RANS or

LES is applied. A more detailed explanation of DES and DDES models can be found in Xing et

al. (2007) and Xing et al. (2010) respectively, while LES options are described in Li et al. (2020).

3.1.4. Overset grids and motions

The computation domain is discretized using curvilinear multiblock structured grids. Large

amplitude motions and control surface deflections are calculated using fully dynamic overset

connectivity. Cartesian local refinements are used to improve the resolution of the detached flow

and to ensure adequate connectivity between grids. The run-time overset domain connectivity

information (DCI) is obtained using the code SUGGAR++ (Noack et al. 2009). This allows for

relative motion between grids and modeling of solid bodies including appendages using composite

grid systems. Motions are implemented using a hierarchy of bodies (Carrica et al., 2007b) by

computing rigid body equations for the parent body (submarine, ship) in a six degrees of freedom

20
(6DOF) solver, and using controllers or imposed input motions to allow the appendages (rudders,

propellers, stabilizers, etc.) to move in one degree of freedom with respect to the parent body.

3.1.5. Free Surface Modeling

The free surface is computed by single phase level set approach (Carrica et al. 2007a),

using a SOR (Successive Over-Relaxation) type solver with second order upwind scheme for

convection. REX can compute a two-phase air/water free surface interface using a semi-coupled

method where the water free surface is decoupled from the air computation (Huang et al. 2008).

3.1.6. Ocean Waves

The evaluation of the model is done under an actual wave spectrum of ocean. This needs

the completion of hydrodynamic model construction and validation through comparison with

model tests and CFD simulations to be discussed in §4 and §5. REX has extensive capabilities to

impose linear waves either regular or irregular, unidirectional or multidirectional. Details are not

discussed here, interested readers are referred to Carrica et al. (2008).

The energy density 𝑆 of Brettschneider spectrum in terms of significant wave frequency

𝜔𝑠 and significant wave height 𝐻𝑆 can is as follows.

5 𝜔𝑠4 2 −5𝜔𝑠44
S(𝜔) = 𝐻 𝑒 4𝜔 , 𝑚2 ∙ 𝑠 (8)
16 𝜔 5 𝑆

The nominal values for 𝐻𝑆 and 𝜔𝑆 can be determined for nominal sea states from the

parameters in table 2.

21
Table 2. Nominal Sea State index and corresponding wave frequencies and heights

Sea State 2π/ω𝑠 (s) H𝑠 (m)


2 6.3 0.3
3 7.5 0.9
4 8.8 1.9
5 9.7 3.3
6 12.4 5.0

3.2. Propulsion and Controllers

REX has a variety of native PID-based controllers, and also interfaces with third-party

controllers. Details of the functionality of the controllers implemented in REX can be found in

Carrica et al. (2020). Propellers can be represented in REX using the discrete geometry of the

blades, body force models such as Hough and Ordway (1965), or by coupling a third-party solver,

such as the potential flow solver PUF-14 (Kerwin et al., 1987). Using a potential flow model

significantly reduces computational cost while maintaining good agreement with experimental and

fully discretized results (Martin et al., 2015). Other appendages are always represented as

discretized bodies with 1DoF and different controllers are implemented to perform maneuvers,

control attitude, etc. Finally, controllers that simulate internal components such as trim and ballast

tanks that affect 6DoF variables directly have also been implemented.

3.2.1. Propeller Cruise Control

A PID (Proportional-Integral-Derivative) feedback controller is used to control speed

changing the propeller RPS

𝑡 𝑑(𝑢 − 𝑢𝑡𝑎𝑟𝑔𝑒𝑡 )
𝑛(𝑡) = 𝐾𝑝 (𝑢 − 𝑢𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐾𝐼 ∫ (𝑢 − 𝑢𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐾𝐷 (9)
0 𝑑𝑡

22
where 𝑛(𝑡) is the propeller rotation rate at current time step 𝑡 (in revolutions per second or RPS),

𝑢 − 𝑢𝑡𝑎𝑟𝑔𝑒𝑡 is the error between the current speed and the target speed, and 𝐾𝑃 , 𝐾𝐼 and 𝐾𝐷 are

proportional, integral and derivative gains.

3.2.2. Vertical and Horizontal Controller

Unlike surface ships for which the attitude is passively maintained, the position of a

submarine relies heavily on active controlling since the entire hull is submerged and remains

approximately neutrally buoyant during operation. For a submarine it is often desired to minimize

pitch angle, rate and acceleration, and control depth and yaw while following desired trajectories.

To achieve these goals, vertical and horizontal controllers following Overpelt et al. (2015) are

implemented as

𝑡 𝑑(𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 )
𝛿𝑉 = 𝑃𝑧 (𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝑧 ∫ (𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐷𝑍
0 𝑑𝑡

𝑡 𝑑(𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 )
+ 𝑃𝜃 (𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝜃 ∫ (𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐷𝜃
0 𝑑𝑡

𝑡 𝑑(𝑦 − 𝑦𝑡𝑎𝑟𝑔𝑒𝑡 )
𝛿𝐻 = 𝑃𝑦 (𝑦 − 𝑦𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝑦 ∫ (𝑦 − 𝑦𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐷𝑦 𝛿𝐻 (10)
0 𝑑𝑡

𝑡
+ 𝑃𝜓 (𝜓 − 𝜓𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝜓 ∫ (𝜓 − 𝜓𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡
0

𝑑(𝜓 − 𝜓𝑡𝑎𝑟𝑔𝑒𝑡 )
+ 𝐷𝜓
𝑑𝑡

where 𝛿𝑉 and 𝛿𝐻 are the vertical and horizontal control command angles, respectively. 𝑃𝑧 , 𝐼𝑧 and

𝐷𝑧 are the proportional, integral and derivative constants for depth, and 𝑃𝜃 , 𝐼𝜃 and 𝐷𝜃 are the

constants for pitch.

23
3.2.3. Trim and Ballast Tank

The trim and ballast tanks control the location of 𝐿𝐶𝐺 and the weight 𝑊 of the submarine,

respectively. They can be selectively used for near surface maneuvering that requires demanding

forces and moments for control. It is inefficient to use them throughout the mission since their

operation requires extra power consumption and generates undesired noise and vibration.

They are modeled following the formulation from Carrica et al. (2016)

𝑡 𝑑(𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 )
Δ𝑚 = 𝑃𝑚 (𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝑚 ∫ (𝑧 − 𝑧𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐷𝑚
0 𝑑𝑡
(11)
𝑡 𝑑(𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 )
Δ𝐿𝐶𝐺 = 𝑃𝐿𝐶𝐺 (𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 ) + 𝐼𝐿𝐶𝐺 ∫ (𝜃 − 𝜃𝑡𝑎𝑟𝑔𝑒𝑡 )𝑑𝑡 + 𝐷𝐿𝐶𝐺
0 𝑑𝑡

Where 𝑃𝑚 , 𝐼𝑚 and 𝐷𝑚 are 𝑃𝐼𝐷 constants for mass change Δ𝑚 and 𝑃𝐿𝐶𝐺 , 𝐼𝐿𝐶𝐺 and 𝐷𝐿𝐶𝐺 are

the 𝑃𝐼𝐷 constants for change in the location of 𝐿𝐶𝐺.

3.3. Computational Grid and Boundary Conditions

3.3.1. Grids

Details of the grid system are shown in table 3. Grids with and without discretized

propeller are used. The presence of a rotating propeller grid is aimed for investigation of side

forces and moments which are not captured by actuator disc models. Refinement grids are added

where needed to resolve specific small scale flow regions.

A refinement grid is added at the free surface as recommended by the ITTC procedure

(2011). A propeller refinement grid is added to capture the flow field around the propeller and its

wake. For captive model tests, the discretized propeller grid is only used when investigating

propeller/hull and stern planes interaction, but the propeller refinement is retained to resolve the

hull and appendage wakes when using an actuator disk model.

24
The computational domain is illustrated in the Figure 4. The Background grid is extended

one body length in each direction of the submarine in the horizontal plane to minimize wave

reflection effects at the boundaries of the computation domain, and to allow imposition of

undisturbed flow boundary conditions. The total height of the background grid is approximately

1.66𝐿0 where the top and bottom surfaces are extended by 0.16𝐿0 and 1.5𝐿0 , respectively, with

respect to the free surface level located at 𝑧 = 0.

Figure 4. Computational domain and boundary conditions

25
3.3.2. Boundary Conditions

The imposed boundary conditions are summarized in table 4. The wall boundaries for the

solid surface of the submarine body uses non-slip wall with wall function (#24) suited for full scale

Reynolds number order of 108 . The background grid uses inlet (#10) and exit (#11), zero-gradient

(#40), far-field (#12, #13) as described in Figure 3.

26
Table 3. Details of computational grids

No. Group Part Name Case I J K Subtotal


1 Hull 177 36 128 815,616
2 Nose 37 22 36 29,304
3 Sail SB 71 22 36 56,232
4 Sail PS 71 22 36 56,232
5 Sail Fin Root SB 107 36 36 138,672
6 Sail Fin Root PS 107 36 36 138,672
7 Stern Plane Root Lower SB 1 43 22 31 29,326
Hull
8 Stern Plane Root Lower SB 2 43 22 31 29,326
9 Stern Plane Root Upper SB 1 43 22 31 29,326
10 Stern Plane Root Upper SB 2 43 22 31 29,326
11 Stern Plane Root Upper PS 1 43 22 31 29,326
12 Stern Plane Root Upper PS 2 43 22 31 29,326
13 Stern Plane Root Lower PS 1 43 22 31 29,326
All Cases
14 Stern Plane Root Lower PS 2 43 22 31 29,326
15 Sail Plane SB1 45 31 60 83,700
16 Sail Plane SB 2 45 31 60 83,700
Sail Planes
17 Sail Plane PS 1 45 31 60 83,700
18 Sail Plane PS 2 45 31 60 83,700
19 Stern Plane SB Lower 1 43 31 43 57,319
20 Stern Plane SB Lower 2 43 31 43 57,319
21 Stern Plane SB Upper 1 43 31 43 57,319
22 Stern Plane SB Upper 2 43 31 43 57,319
Stern Planes
23 Stern Plane PS Upper 1 43 31 43 57,319
24 Stern Plane PS Upper 2 43 31 43 57,319
25 Stern Plane PS Lower 1 43 31 43 57,319
26 Stern Plane PS Lower 2 43 31 43 57,319
27 Propeller Hub 51 22 101 113,322
28 Blade 1 71 22 36 56,232
29 Tip 1 37 22 36 29,304
30 Blade 2 71 22 36 56,232
31 Tip 2 37 22 36 29,304
32 Blade 3 71 22 36 56,232
Propeller
33 Propeller Tip 3 37 22 36 29,304
Interaction
34 Blade 4 71 22 36 56,232
Tests
35 Tip 4 37 22 36 29,304
36 Blade 5 71 22 36 56,232
37 Tip 5 37 22 36 29,304
38 Blade 6 71 22 36 56,232
39 Tip 6 37 22 36 29,304
40 Propeller Refinement 107 71 64 486,208
41 Refinements General Refinement 3 2 1 486,208
All Cases
42 Sail Wake Refinement 3 2 1 486,208
43 Stern Plane Refinement 3 2 1 486,208
44 Free Surface Refinement 4 2 1 657,943
45 Background 3 1 3 732,736

27
Table 4. Details of boundary conditions

 p k  U V W
p
9  = −z =0 k fs = 10−7  fs = 9 U = ufullspd V = vfullspd W = wfullspd
inlet n
10
Far inlet
 = −z p=0 k fs = 10−7  fs = 9 U = ufullspd V = vfullspd W = wfullspd

11  p k   2U  2V  2W
=0 =0 =0 =0 =0 =0 =0
exit n n n n  n2  n2  n2
12 k  U V W
Very far-  =0 p=0 =0 =0 =0 =0 =0
field n n n n n
13  p k  U V W
=0 =0 =0 =0 =0 =0 =0
far-field n n n n n n n
24  Not Wall Wall
no slip wall =0 Wall function Wall function Wall function
function n needed function
function
40  p k  U V W
zero =0 =0 =0 =0 =0 =0 =0
gradient n n n n n n n

28
4. DYNAMIC MODEL OF A SUBMARINE

4.1. Rigid Body Motion

A force balance in the body fixed local coordinate system described in figure 2 gives

𝑴 ∙ 𝒔̇ = 𝑭 − 𝒃 (12)

where 𝑀 is the mass matrix, 𝑠 is the velocity vector, 𝐹 is the external loads vector, and 𝑏 is the

inertial coupling vector which arises. These are defined as

𝒔 = [𝑢, 𝑣, 𝑤, 𝑝, 𝑞, 𝑟]𝑇 (13)

𝑚 0 0 0 𝑚𝑧𝐺 −𝑚𝑦𝐺
0 𝑚 0 −𝑚𝑧𝐺 0 𝑚𝑥𝐺
0 0 𝑚 𝑚𝑦𝐺 −𝑚𝑥𝐺 0
𝑴= 0 −𝑚𝑧𝐺 𝑚𝑦𝐺 𝐼𝑥𝑥 −𝐼𝑥𝑦 −𝐼𝑥𝑧 (14)
𝑚𝑧𝐺 0 −𝑚𝑥𝐺 −𝐼𝑥𝑦 𝐼𝑦𝑦 −𝐼𝑦𝑧
[−𝑚𝑦𝐺 𝑚𝑧𝐺 0 −𝐼𝑥𝑧 −𝐼𝑦𝑧 𝐼𝑧𝑧 ]

−(𝑊 − 𝐵) sin 𝜃 + 𝑋𝐻 (𝑡, 𝑢, 𝑢̇ )


(𝑊 − 𝐵) cos 𝜃 sin 𝜙 + 𝑌𝐻 (𝑡, 𝑢, 𝑢̇ )
(𝑊 − 𝐵) cos 𝜃 cos 𝜙 + 𝑍𝐻 (𝑡, 𝑢, 𝑢̇ )
𝑭= (15)
(𝑦𝐺 𝑊 − 𝑦𝐵 𝐵) cos 𝜃 cos 𝜙 − (𝑧𝐺 𝑊 − 𝑧𝐵 𝑊) cos 𝜃 sin 𝜙 + 𝐾𝐻 (𝑡, 𝑢, 𝑢̇ )
−(𝑥𝐺 𝑊 − 𝑥𝐵 𝐵) cos 𝜃 cos 𝜙 − (𝑧𝐺 𝑊 − 𝑧𝐵 𝐵) sin 𝜃 + 𝑀𝐻 (𝑡, 𝑢, 𝑢̇ )
[ −(𝑥𝐺 𝑊 − 𝑥𝐵 𝐵) cos 𝜃 sin 𝜙 + (𝑦𝐺 𝑊 − 𝑦𝐵 𝐵) sin 𝜃 + 𝑁𝐻 (𝑡, 𝑢, 𝑢̇ ) ]

29
𝑚[𝑤𝑞 − 𝑣𝑟 − 𝑥𝐺 (𝑞 2 + 𝑟 2 ) + 𝑦𝐺 (𝑝𝑞 − 𝑟̇ ) + 𝑧𝐺 (𝑝𝑟 + 𝑞̇ )]
𝑚[𝑢𝑟 − 𝑤𝑝 − 𝑦𝐺 (𝑟 2 + 𝑝2 ) + 𝑧𝐺 (𝑞𝑟 − 𝑝̇ ) + 𝑥𝐺 (𝑞𝑝 + 𝑟̇ )]
𝑚[𝑣𝑝 − 𝑢𝑞 − 𝑧𝐺 (𝑝2 + 𝑞 2 ) + 𝑥𝐺 (𝑟𝑝 − 𝑞̇ ) + 𝑦𝐺 (𝑟𝑞 + 𝑝̇ )]
𝒃= (16)
(𝐼𝑧 − 𝐼𝑦 )𝑞𝑟 + 𝑚[𝑦𝐺 (𝑤̇ − 𝑢𝑞 + 𝑣𝑝) − 𝑧𝐺 (𝑣̇ − 𝑤𝑝 + 𝑢𝑟)]
(𝐼𝑥 − 𝐼𝑧 )𝑟𝑝 + 𝑚[𝑧𝐺 (𝑢̇ − 𝑣𝑟 + 𝑤𝑞) − 𝑥𝐺 (𝑤̇ − 𝑢𝑞 + 𝑣𝑝)]
[(𝐼𝑦 − 𝐼𝑥 )𝑝𝑞 + 𝑚[𝑥𝐺 (𝑣̇ − 𝑤𝑝 + 𝑢𝑟) − 𝑦𝐺 (𝑢̇ − 𝑣𝑟 + 𝑤𝑞)]]

The six components of the external load vector are axial, lateral and vertical forces, and

rolling, pitching and yawing moments, in that order. Each has contributions from the submarine

weight 𝑊, buoyancy B, and hydrodynamic forces and moments 𝐹𝐻 = (𝑋𝐻 , 𝑌𝐻 , 𝑍𝐻 , 𝐾𝐻 , 𝑀𝐻 , 𝑁𝐻 ).

The point of application is located at 𝐶𝐺 of the submarine, hence the actual location may change

slightly by the actuation of the trim tanks, which shifts the longitudinal center of gravity 𝐿𝐶𝐺 of

the submarine and also possibly the lateral and vertical locations if the tanks are located at different

heights or lateral positions. Likewise, the weight of submarine may change due to actuation of the

ballast tank. However, the buoyancy and its point of application does not change if the submarine

is intact since it is determined by the displaced volume in the water. The acceleration vector 𝑠̇ can

be obtained implicitly by the 6-DoF equations of motion from Eq. (12). The velocities and

displacement are obtained by temporally integrating the vector 𝑠̇ .

4.2. Hydrodynamic Forces and Moments

4.2.1. Gertler and Hagen Model for X shape Stern Planes

Accurate predictions of rigid body motions require precise modeling of hydrodynamic

loads. The hydrodynamic forces and moments are a function velocity, acceleration and orientation

of the body. Gertler and Hagen (1967) proposed a model based mostly on steady state assumptions

with a few unsteady terms to describe the effects of accelerations. The hydrodynamic forces and

30
moments are described as function of velocities and accelerations of the moving body. Notice that

the acceleration appears in both rigid body equations and hydrodynamic loading terms. In

coefficient-based models that cannot separate the unsteady hydrodynamic loading into linear

components it is not possible to discretize explicitly in time. Watt (2007) extracted acceleration

terms and used ordinary discretization by describing the loading using the concept of added mass.

Bettle 2013 evaluated coefficients in the model with steady state CFD simulations and did not

explicitly describe the terms, therefore solved the equations iteratively using a predictor-corrector

integration scheme.

The Gertler and Hagen (1967) model for hydrodynamic loads is modified for X-shape stern

plane as follows.

1 4 ′ 2 1
𝑋𝐻 = ′ 2
𝜌𝐿 [𝑋𝑞𝑞 𝑞 + 𝑋𝑟𝑟 ′
𝑟 + 𝑋𝑟𝑝 𝑟𝑝] + 𝜌𝐿3 [𝑋𝑢̇′ 𝑢̇ + 𝑋𝑣𝑟
′ ′
𝑣𝑟 + 𝑋𝑤𝑞 𝑤𝑞]
2 2
1
+ 𝜌𝐿2 [𝑋𝑢𝑢

𝑢2 + 𝑋𝑣𝑣

𝑣 2 + 𝑋𝑤𝑤

𝑤 2] (17)
2
4
1 ′ ′
+ 𝜌𝐿2 [∑ 𝑋𝛿𝑋 𝑖 𝛿𝑋𝑖
𝑢2 𝛿𝑋𝑖2 + 𝑋𝛿𝐵𝛿𝐵 𝑢2 𝛿𝑋𝐵2 ] + 𝑇𝑝𝑟𝑜𝑝
2
𝑖=1

1 4 ′ ′
𝑌𝐻 = 𝜌𝐿 [𝑌𝑟̇ 𝑟̇ + 𝑌𝑝̇′ 𝑝̇ + 𝑌𝑝|𝑝| ′
𝑝|𝑝| + 𝑌𝑝𝑞 ′
𝑝𝑞 + 𝑌𝑞𝑟 𝑞𝑟]
2
1 ′ 𝑣 1
+ 𝜌𝐿3 [𝑌𝑣̇′ 𝑣̇ + 𝑌𝑟′ 𝑢𝑟 + 𝑌𝑝′ 𝑢𝑝 + 𝑌𝑣|𝑟| |(𝑣 2 + 𝑤 2 )2 | |𝑟| + 𝑌𝑣𝑞

𝑣𝑞
2 |𝑣|
4
′ ′
1 ′
+ 𝑌𝑤𝑝 𝑤𝑝 + 𝑌𝑤𝑟 𝑤𝑟] + 𝜌𝐿3 ∑ 𝑌|𝑟|𝑋 𝑢|𝑟|𝛿𝑋𝑖 (18)
2 𝑖
𝑖=1

1 ′
1
+ 𝜌𝐿2 [𝑌∗′ 𝑢2 + 𝑌𝑣′ 𝑢𝑣 + 𝑌𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 | + 𝑌𝑣𝑤

𝑣𝑤]
2
4
1 ′
+ 𝜌𝐿2 ∑ 𝑌𝛿𝑋 𝑢2 𝛿𝑋𝑖2 + 𝑌𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

31
1
𝑍𝐻 = 𝜌𝐿4 [𝑍𝑞̇′ 𝑞̇ + 𝑍𝑝𝑝

𝑝2 + 𝑍𝑟𝑟
′ 2 ′
𝑟 + 𝑍𝑟𝑝 𝑟𝑝]
2
1 ′ ′ 𝑤 1
+ 𝜌𝐿3 [𝑍𝑤 ′ ′ ′
̇ 𝑤̇ + 𝑍𝑣𝑟 𝑣𝑟 + 𝑍𝑣𝑝 𝑣𝑝 + 𝑍𝑞 𝑢𝑞 + 𝑍𝑤|𝑞| |(𝑣 2 + 𝑤 2 )2 | |𝑞|]
2 |𝑤|
4
1 ′
+ 𝜌𝐿3 ∑ 𝑍|𝑞|𝛿 𝑢|𝑞|𝛿𝑋𝑖 (19)
2 𝑋𝑖
𝑖=1

1 ′
1
+ 𝜌𝐿2 [𝑍∗′ 𝑢2 + 𝑍𝑣𝑣

𝑣 2 + 𝑍𝑣′ 𝑢𝑣 + 𝑍𝑤

𝑢𝑤 + 𝑍𝑤|𝑤| 𝑤 |(𝑣 2 + 𝑤 2 )2 |]
2
4
𝜌 ′ ′
+ 𝐿2 𝑢2 [∑ 𝑍𝛿𝑋 2 2
𝛿𝑋𝑖 𝑢 𝛿𝑋𝑖 + 𝑍𝛿𝐵 𝑢 𝛿𝐵 ] + 𝑍𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

1 5 ′ ′
𝐾𝐻 = 𝜌𝐿 [𝐾𝑝̇ 𝑝̇ + 𝐾𝑟̇′ 𝑟̇ + 𝐾𝑞𝑟
′ ′
𝑞𝑟 + 𝐾𝑝𝑞 𝑝𝑞 + 𝐾𝑝|𝑝| 𝑝|𝑝|]
2
1
+ 𝜌𝐿4 [𝐾𝑣̇′ 𝑣̇ + 𝐾𝑝′ 𝑢𝑝 + 𝐾𝑟′ 𝑢𝑟 + 𝐾𝑣𝑞
′ ′
𝑣𝑞 + 𝐾𝑤𝑝 ′
𝑤𝑝 + 𝐾𝑤𝑟 𝑤𝑟 ]
2
1 1 (20)

+ 𝜌𝐿3 [𝐾∗′ 𝑢2 + 𝐾𝑣′ 𝑢𝑣 + 𝐾𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 | + 𝐾𝑣𝑤

𝑣𝑤]
2
4
1 ′
+ 𝜌𝐿3 ∑ 𝐾𝛿𝑋 𝑖 𝛿𝑋𝑖
𝑢2 𝛿𝑋𝑖2 + 𝐾𝑝𝑟𝑜𝑝
2
𝑖=1

32
1 ′
𝑀𝐻 = 𝜌𝐿5 [𝑀𝑞̇′ 𝑞̇ + 𝑀𝑝𝑝

𝑝2 + 𝑀𝑟𝑟
′ 2 ′
𝑟 + 𝑀𝑟𝑝 𝑟𝑝 + 𝑀𝑞|𝑞| 𝑞|𝑞|]
2
1 ′ ′
1
+ 𝜌𝐿4 [𝑀𝑤 ′ ′ ′ 2 2
̇ 𝑤̇ + 𝑀𝑣𝑟 𝑣𝑟 + 𝑀𝑣𝑝 𝑣𝑝 + 𝑀𝑞 𝑢𝑞 + 𝑀|𝑤|𝑞 |(𝑣 + 𝑤 )2 | 𝑞]
2
4 4
1 ′ ′
+ 𝜌𝐿4 [∑ 𝑀|𝑞|𝛿𝑋 𝑢|𝑞|𝛿𝑋𝑖 + ∑ 𝑀|𝑟|𝛿𝑋 𝑢|𝑟|𝛿𝑋𝑖 ]
2 𝑖 𝑖
𝑖=1 𝑖=1
(21)
1 ′ ′
1
+ 𝜌𝐿3 [𝑀∗′ 𝑢2 + 𝑀𝑣𝑣

𝑣 2 + 𝑀𝑤

𝑢𝑤 + 𝑀|𝑤| 𝑢|𝑤| + 𝑀𝑤|𝑤| 𝑤 |(𝑣 2 + 𝑤 2 )2 |
2
4
1 1 ′ ′
+ ′
𝑀𝑤𝑤 |𝑤(𝑣 + 𝑤2 2 )2
|] + 𝜌𝐿3 [∑ 𝑀𝛿𝑋 𝑖 𝛿𝑋𝑖
𝑢2 𝛿𝑋𝑖2 + 𝑀𝛿𝐵𝛿𝐵 𝑢2 𝛿𝑋𝐵2 ]
2
𝑖=1

+ 𝑀𝑝𝑟𝑜𝑝

1 ′
𝑁𝐻 = 𝜌𝐿5 [𝑁𝑟̇′ 𝑟̇ + 𝑁𝑝̇′ 𝑝̇ + 𝑁𝑝𝑞
′ ′
𝑝𝑞 + 𝑁𝑞𝑟 𝑞𝑟 + 𝑁𝑟|𝑟| 𝑟|𝑟|]
2
1
+ 𝜌𝐿4 [𝑁𝑣̇′ 𝑣̇ + 𝑁𝑤𝑟
′ ′
𝑤𝑟 + 𝑁𝑤𝑝 ′
𝑤𝑝 + 𝑁𝑣𝑞 𝑣𝑞 + 𝑁𝑝′ 𝑢𝑝 + 𝑁𝑟′ 𝑢𝑟
2
4
′ 2
1
2 )2
1 ′
+ 𝑁|𝑣|𝑟 |(𝑣 + 𝑤 | 𝑟] + 𝜌𝐿4 ∑ 𝑁|𝑟|𝛿𝑋 𝑢|𝑟|𝛿𝑋𝑖
2 𝑖 (22)
𝑖=1

1 ′
1
+ 𝜌𝐿3 [𝑁∗′ 𝑢2 + 𝑁𝑣′ 𝑢𝑣 + 𝑁𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 | + 𝑁𝑣𝑤

𝑣𝑤]
2
4
1 ′
+ 𝜌𝐿3 ∑ 𝑁𝛿𝑋 𝑖 𝛿𝑋𝑖
𝑢2 𝛿𝑋𝑖2 + 𝑁𝑝𝑟𝑜𝑝
2
𝑖=1

Note that the original Gertler and Hagen (1967) model is formulated for a cruciform stern

plane configuration. Modifications are made to obtain the model equations for X-shape stern

planes, including high-order corrections. The revised model can also be used for the cruciform

stern planes, but it requires several stern plane coefficients to be zero. The original model assumes

that the horizontal planes do not contribute to horizontal plane maneuvers and vertical planes do

33
not have authority in vertical plane maneuvers; coefficients need to be truncated depending on

how the plane indices are assigned.

The hydrodynamic forces in 𝑋𝐻 arise from rotational motions, coupled linear and rotational

motions, linear motions with incidence angle, control plane deflection and propeller thrust. 𝑋𝑞𝑞


and 𝑋𝑟𝑟 are respectively due to added mass induced by pitching and yawing motion of the
′ ′ ′
submarine. While 𝑋𝑟𝑟 is symmetric, 𝑋𝑞𝑞 is not due to the hull shape asymmetry to pitch. 𝑋𝑟𝑝

arises from the submarine rolling and yawing at the same time, particularly in horizontal plane
′ ′
maneuvers. 𝑋𝑣𝑟 and 𝑋𝑤𝑞 arise from coupled sway-yaw and heave-pitch respectively and both are

sign specific. 𝑋𝑢̇′ is the added resistance from added mass due to forward acceleration. 𝑋𝑢𝑢
′ ′
, 𝑋𝑣𝑣


and 𝑋𝑤𝑤 describe the body drift force due to incidence angle. Their sum always acts as major

source of resistance; however it is non-symmetric about the vertical drift due to the presence of
′ ′
the sail. 𝑋𝛿𝑋 𝑖 𝛿𝑋𝑖
and 𝑋𝛿𝐵𝛿𝐵 are resistance due to the deflection of each stern plane and of the sail

planes, respectively. 𝑇𝑝𝑟𝑜𝑝 is the propeller thrust as modeled in §3-4.

The hydrodynamic sway force 𝑌𝐻 can be interpreted in similar fashion. The terms related

to the motions occurring in the horizontal plane of the submarine are symmetric except for the

propeller lateral force. 𝑌𝑝̇′ , 𝑌𝑟̇′ and 𝑌𝑣̇′ are added mass terms in y-direction due to acceleration of

rolling, yawing and sway motions. The sway added mass due to pitching is neglected because of

the symmetry of the hull shape about the vertical plane. Rotating motions besides roll are assumed

to induce significant forces in y-direction only when coupled with other rotations or translational
′ ′
velocities. 𝑌𝛿𝑋 𝑖
and 𝑌|𝑟|𝑋𝑖
𝑢|𝑟| are the lateral control force coefficients from the stern plane

deflections and the higher-order correction terms due to yawing motions. The sail planes are not

included since they move synchronously and compensate the lateral force from each other (except

34

for neglected higher-order terms like lateral sail plane forces while turning). 𝑌∗′ , 𝑌𝑣′ , 𝑌𝑣|𝑣| and


𝑌𝑣𝑤 𝑣𝑤 arise due to flow incidence, where 𝑌∗′ is the lateral force coefficient with zero incidence.
′ ′
𝑍𝑤 ̇ and 𝑍𝑞̇ are added mass coefficients due to accelerating heave and pitch motions.

Contributions to vertical added mass from accelerations in other directions are not considered
′ ′
significant. 𝑍𝑝𝑝 , 𝑍𝑟𝑟 are the vertical force coefficients due to pure rolling and yawing. Pure

pitching is not considered to cause significant vertical force. For axially symmetric bodies, there
′ ′
is no coupling for vertical force to sway and rolling. 𝑍𝑣𝑟 and 𝑍𝑣𝑝 arise when sway velocity is


coupled to rolling and yawing. 𝑍∗′ , 𝑍𝑣𝑣

, 𝑍𝑣′ , 𝑍𝑤

and 𝑍𝑤|𝑤| appear as body drift force terms. As in

Eq. (18), 𝑍∗′ is the vertical force coefficient when there is no incidence angle of the flow. The drift

force term is extremely important for vertical plane maneuvers, since the maneuverability of a

submarine depends on the directional stability of the hull. An increase in vertical force causes

vertical acceleration, hence a vertical drift angle. This affects the overall maneuverability in the

vertical plane by altering the pitch of the hull. The damping from 𝑍𝑞′ and 𝑍𝑤|𝑞| affects the vertical


drift angle in a similar way. The vertical forces from sail and stern planes are accounted by 𝑍𝛿𝑋 𝑖 𝛿𝑋𝑖

′ ′
and 𝑍𝛿𝐵 . The stern plane force correction due to pitching motion is reflected by 𝑍|𝑞|𝛿𝑋
, however
𝑖

it is not considered for the sail planes because the pitch rate does not cause change in longitudinal

or vertical flow speed and angle of attack due to short distance from the 𝐶𝐺.

𝐾𝑝̇′ and 𝐾𝑝|𝑝| appear in the hydrodynamic rolling moment equation as damping terms due

to the added mass in roll acceleration and the resistance to the rolling motion. The sign always

opposes rotation. 𝐾𝑟̇′ and 𝐾𝑣̇′ are added mass components due to acceleration in roll and sway,

respectively, and may either increase or decrease the rolling moment depending on the roll stability
′ ′
of the submarine. 𝐾𝑞𝑟 and 𝐾𝑝𝑞 are the components arising when pitching is coupled with yawing

35
and rolling, respectively. The derivatives accounting for pure yawing 𝐾𝑟′ and yawing-rolling 𝐾𝑝𝑟

are not considered in the equation due to weak coupling between them. The components

accounting for coupling between translation and rotation appear in the second bracket in the right-

hand side of Eq. (20), where 𝐾𝑝′ and 𝐾𝑤𝑝



are damping terms. The rolling moment due to drift


angles is accounted by 𝐾∗′ , 𝐾𝑣′ , 𝐾𝑣|𝑣| ′
and 𝐾𝑣𝑤 . The rolling moment due to stern plane deflections is


considered by 𝐾𝛿𝑋 𝑖 𝛿𝑋𝑖
, however higher-order corrections of this moment are neglected. The

contribution of the sail planes is not included for the same reason as the sway force Eq. (18).

The terms 𝑀𝑞̇′ 𝑞̇ , 𝑀𝑤



̇ in the Eq. (21) account for the contribution of added mass due to

acceleration. The damping moment due to change in orientation in the vertical plane is accounted
′ ′
by 𝑀𝑞|𝑞| , 𝑀|𝑤|𝑞 and 𝑀𝑞′ . Similar to damping by heave and pitch in vertical force, the amount of

damping produced influences the vertical plane maneuverability. Since the body is directionally

unstable, the body drift moment accounted by the sum of components with 𝑀∗′ , 𝑀𝑣𝑣
′ ′
, 𝑀𝑤 , 𝑀|𝑤| and


𝑀𝑤|𝑤| tend to increase in the same direction as the orientation of the hull. When this moment is

too large to be dampened, the control surfaces will not be able to control the orientation of the
′ ′
submarine. The body drift, damping , and control moments computed with 𝑀𝛿𝑋 𝑖 𝛿𝑋𝑖
, 𝑀𝛿𝐵𝛿𝐵

′ ′
𝑀|𝑞|𝛿𝑋 𝑖
and 𝑀|𝑟|𝛿𝑋 𝑖
should be precisely evaluated for good prediction of vertical plane maneuvers.


One of the higher-order corrections for stern planes 𝑀|𝑟|𝛿𝑋 𝑖
is neglected in the formulation for

cruciform stern planes since the horizontal stern planes are not significantly affected by the yaw

motions, however it should be included for X stern planes as all planes produce vertical and

horizontal control forces and moments at any deflections.

The added mass affecting the yawing moment in Eq. (22) arises from acceleration in yaw

and sway in 𝑁𝑟̇′ and 𝑁𝑣̇′ . The horizontal plane maneuverability can be understood in similar manner

36

as for vertical plane maneuvers. The drift moment appears as the sum of the terms 𝑁∗′ , 𝑁𝑣′ , 𝑁𝑣|𝑣|

′ ′
and 𝑁𝑣𝑤 and will accelerate yawing motions, while damping moments from 𝑁𝑟|𝑟| 𝑟|𝑟|, 𝑁𝑟′ 𝑢𝑟 and


𝑁|𝑣|𝑟 oppose yawing.

For all equations, the in-plane components have larger authority to determine the behavior

of motion, as the off-plane components affect the tendency of submarine to maintain the current

maneuvering plane. They usually have very small magnitude and the test results are hard to

extrapolate, therefore careful decision of test cases is required.

4.2.2. Model Reduction by Simplified High Order Correction

The model equation includes several high-order coupling terms. They appear in the
′ ′ ′ ′
equations as 𝑌|𝑟|𝑋 𝑖
, 𝑍|𝑞|𝛿𝑋
, 𝑀|𝑞|𝛿𝑋 𝑖
and 𝑁|𝑟|𝛿𝑋 𝑖
which account for the variance of controlling
𝑖

performance of stern planes due to rotational motions. This is because the stern planes are located

far from the center of gravity, the rate of rotational motion can affect the actual angle of attack to

the planes significantly. The high-order correction terms for stern planes are expressed as
4 4 4
1 3 ′ 1 ′ 2
1 2 ′
𝜌𝐿 ∑ 𝑌|𝑟|𝑋 𝑢|𝑟|𝛿𝑋𝑖 + 𝜌𝐿2 ∑ 𝑌𝛿𝑋 2
𝛿𝑋𝑖 𝑢 𝛿𝑋𝑖 = 𝜌𝐿 ∑ 𝑌𝛿𝑋𝑖 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢
2
2 𝑖 2 𝑖 2
𝑖=1 𝑖=1 𝑖=1

4 4 4
1 3 ′ 1 2 ′ 2 2
1 2 ′
𝜌𝐿 ∑ 𝑍|𝑞|𝛿 𝑢|𝑞|𝑋𝑖 + 𝜌𝐿 ∑ 𝑍𝛿𝑋 𝑢 𝛿𝑋𝑖 = 𝜌𝐿 ∑ 𝑍𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2
2 𝑋𝑖 2 𝑖 2 𝑖
𝑖=1 𝑖=1 𝑖=1
(23)
4 4 4
1 4 ′ 1 ′ 1 ′
𝜌𝐿 ∑ 𝑀|𝑞|𝛿𝑋 𝑢|𝑞|𝛿𝑋𝑖 + 𝜌𝐿3 ∑ 𝑀𝛿𝑋 𝑢2 𝛿𝑋𝑖2 = 𝜌𝐿3 ∑ 𝑀𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2
2 𝑖 2 𝑖 2 𝑖
𝑖=1 𝑖=1 𝑖=1

4 4 4
1 4 ′ 1 ′ 2
1 3 ′
𝜌𝐿 ∑ 𝑁|𝑟|𝛿𝑋 𝑢|𝑟|𝛿𝑋𝑖 + 𝜌𝐿3 ∑ 𝑁𝛿𝑋 2
𝛿𝑋𝑖 𝑢 𝛿𝑋𝑖 = 𝜌𝐿 ∑ 𝑁𝛿𝑋𝑖 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢
2
2 𝑖 2 𝑖 2
𝑖=1 𝑖=1 𝑖=1

In equation (23), 𝛿𝑋𝑖,𝑒𝑓𝑓 is the effective angle of attack to the 𝑖th stern plane corrected by

the orientation and the position with respect to the center of gravity; detailed computation is

37
discussed in §4.2.3. The sail plane performance may also be written in terms of the effective angle

of attack to the sail plane 𝛿𝑋𝐵,𝑒𝑓𝑓 as

′ ′
𝑋𝛿𝐵𝛿𝐵 𝑢2 𝛿𝑋𝐵2 = 𝑋𝛿𝐵 (𝛿𝑋𝐵,𝑒𝑓𝑓 , 𝑧)𝑢2

′ ′
𝑍𝛿𝐵𝛿𝐵 𝑢2 𝛿𝑋𝐵2 = 𝑍𝛿𝐵 (𝛿𝑋𝐵,𝑒𝑓𝑓 , 𝑧)𝑢2 (24)

′ ′
𝑀𝛿𝐵𝛿𝐵 𝑢2 𝛿𝑋𝐵2 = 𝑀𝛿𝐵 (𝛿𝑋𝐵,𝑒𝑓𝑓 , 𝑧)𝑢2

The surge, sway and heave velocities can be expressed alternatively by the total velocity

𝑈 = √𝑢2 + 𝑣 2 + 𝑤 2 , vertical drift angle 𝛼 and the horizontal drift 𝛽. The loading terms due to

translational velocities can be substituted to an alternative form

𝑋 ′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 = 𝑋𝑢𝑢



𝑢2 + 𝑋𝑣𝑣

𝑣 2 + 𝑋𝑤𝑤

𝑤2
1

𝑌 ′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 = 𝑌∗′ 𝑢2 + 𝑌𝑣′ 𝑢𝑣 + 𝑌𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 | + 𝑌𝑣𝑤

𝑣𝑤
1

𝑍 ′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 = 𝑍∗′ 𝑢2 + 𝑍𝑣′ 𝑢𝑣 + 𝑍𝑤

𝑢𝑤 + 𝑍𝑤|𝑤| 𝑤 |(𝑣 2 + 𝑤 2 )2 |
1

𝐾 ′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 = 𝐾∗′ 𝑢2 + 𝐾𝑣′ 𝑢𝑣 + 𝐾𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 | + 𝐾𝑣𝑤

𝑣𝑤
(25)

𝑀′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 = 𝑀∗′ 𝑢2 + 𝑀𝑣𝑣

𝑣 2 + 𝑀𝑤

𝑢𝑤 + 𝑀|𝑤| 𝑢|𝑤|
1 1

+ 𝑀𝑤|𝑤| 𝑤 |(𝑣 2 + 𝑤 2 )2 | + 𝑀𝑤𝑤

|𝑤(𝑣 2 + 𝑤 2 )2 |
1

𝑁′(𝑈, 𝛼, 𝛽, 𝑧) 𝑈 2 = 𝑁∗′ 𝑢2 + 𝑁𝑣′ 𝑢𝑣 + 𝑁𝑣|𝑣| 𝑣 |(𝑣 2 + 𝑤 2 )2 |

+ 𝑁𝑣𝑤 𝑣𝑤

The simplified form of the mathematical model for hydrodynamic forces and moments

after substituting the given relations is

38
1 4 ′ 2 1
𝑋𝐻 = ′ 2
𝜌𝐿 [𝑋𝑞𝑞 𝑞 + 𝑋𝑟𝑟 ′
𝑟 + 𝑋𝑟𝑝 𝑟𝑝] + 𝜌𝐿3 [𝑋𝑢̇′ 𝑢̇ + 𝑋𝑣𝑟
′ ′
𝑣𝑟 + 𝑋𝑤𝑞 𝑤𝑞]
2 2
1
+ 𝜌𝐿2 𝑋𝑢𝑣𝑤
′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 (26)
2
4
1 ′ ′
+ 𝜌𝐿2 [∑ 𝑋𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2 + 𝑋𝛿𝐵 (𝛿𝑋𝐵,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝑇𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

1 4 ′ ′
𝑌𝐻 = 𝜌𝐿 [𝑌𝑟̇ 𝑟̇ + 𝑌𝑝̇′ 𝑝̇ + 𝑌𝑝|𝑝| ′
𝑝|𝑝| + 𝑌𝑝𝑞 ′
𝑞𝑝 + 𝑌𝑞𝑟 𝑞𝑟]
2
1 ′ 𝑣 1
+ 𝜌𝐿3 [𝑌𝑣̇′ 𝑣̇ + 𝑌𝑝′ 𝑢𝑝 + 𝑌𝑟′ 𝑢𝑟 + 𝑌𝑣𝑞

𝑣𝑞 + 𝑌𝑣|𝑟| |(𝑣 2 + 𝑤 2 )2 | |𝑟|
2 |𝑣|
(27)
′ ′
+ 𝑌𝑤𝑝 𝑤𝑝 + 𝑌𝑤𝑟 𝑤𝑟]

4
1 ′
+ 𝜌𝐿2 [𝑌𝑢𝑣𝑤

(𝑈, 𝛼, 𝛽, 𝑧) 𝑈 2 + ∑ 𝑌𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝑌𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

1
𝑍𝐻 = 𝜌𝐿4 [𝑍𝑞̇′ 𝑞̇ + 𝑍𝑝𝑝

𝑝2 + 𝑍𝑟𝑟
′ 2 ′
𝑟 + 𝑍𝑟𝑝 𝑟𝑝]
2
1 ′ ′ 𝑤 1
+ 𝜌𝐿3 [𝑍𝑤 ̇ 𝑤̇ + 𝑍𝑞

𝑢𝑞 + 𝑍 ′
𝑣𝑝 𝑣𝑝 + 𝑍 ′
𝑣𝑟 𝑣𝑟 + 𝑍𝑤|𝑞| |(𝑣 2
+ 𝑤 2 )2 |𝑞|]
|
2 |𝑤|
4 (28)
1 ′
+ 𝜌𝐿2 [𝑍𝑢𝑣𝑤
′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 + ∑ 𝑍𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2
2 𝑖
𝑖=1


+ 𝑋𝛿𝐵 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝑍𝑝𝑟𝑜𝑝

1 5 ′ ′
𝐾𝐻 = 𝜌𝐿 [𝐾𝑝̇ 𝑝̇ + 𝐾𝑟̇′ 𝑟̇ + 𝐾𝑞𝑟
′ ′
𝑞𝑟 + 𝐾𝑝𝑞 𝑝𝑞 + 𝐾𝑝|𝑝| 𝑝|𝑝|]
2
1
+ 𝜌𝐿4 [𝐾𝑣̇′ 𝑣̇ + 𝐾𝑝′ 𝑢𝑝 + 𝐾𝑟′ 𝑢𝑟 + 𝐾𝑣𝑞
′ ′
𝑣𝑞 + 𝐾𝑤𝑝 ′
𝑤𝑝 + 𝐾𝑤𝑟 𝑤𝑟 ] (29)
2
4
1 ′
+ 𝜌𝐿3 [𝐾𝑢𝑣𝑤 ′(𝑈, 𝛼, 𝛽, 𝑧) 𝑈 2 + ∑ 𝐾𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝐾𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

39
1 5 ′ ′ ′
𝑀𝐻 = 𝜌𝐿 [𝑀𝑞̇ 𝑞̇ + 𝑀𝑝𝑝 𝑝2 + 𝑀𝑟𝑟
′ 2 ′
𝑟 + 𝑀𝑟𝑝 𝑟𝑝 + 𝑀𝑞|𝑞| 𝑞|𝑞|]
2
1 ′ ′
1
+ 𝜌𝐿4 [𝑀𝑤 ′ ′ ′ 2 2
̇ 𝑤̇ + 𝑀𝑞 𝑢𝑞 + 𝑀𝑣𝑝 𝑣𝑝 + 𝑀𝑣𝑟 𝑣𝑟 + 𝑀|𝑤|𝑞 |(𝑣 + 𝑤 )2 | 𝑞]
2
4 (30)
1 ′
+ 𝜌𝐿3 [𝑀𝑢𝑣𝑤
′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 + ∑ 𝑀𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2
2 𝑖
𝑖=1


+ 𝑀𝛿𝐵 (𝛿𝑋𝐵,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝑀𝑝𝑟𝑜𝑝

1 5 ′ ′
𝑁𝐻 = 𝜌𝐿 [𝑁𝑟̇ 𝑟̇ + 𝑁𝑝̇′ 𝑝̇ + 𝑁𝑝𝑞
′ ′
𝑝𝑞 + 𝑁𝑞𝑟 𝑞𝑟 + 𝑁𝑟|𝑟| 𝑟|𝑟|]
2
1 ′
1
+ 𝜌𝐿4 [𝑁𝑣̇′ 𝑣̇ + 𝑁𝑝′ 𝑢𝑝 + 𝑁𝑟′ 𝑢𝑟 + 𝑁𝑣𝑞

𝑣𝑞 + 𝑁|𝑣|𝑟 |(𝑣 2 + 𝑤 2 )2 | 𝑟
2
(31)
′ ′
+ 𝑁𝑤𝑝 𝑤𝑝 + 𝑁𝑤𝑟 𝑤𝑟]

4
1 ′
+ 𝜌𝐿3 [𝑁𝑢𝑣𝑤
′ (𝑈, 𝛼, 𝛽, 𝑧)𝑈 2 + ∑ 𝑁𝛿𝑋 (𝛿𝑋𝑖,𝑒𝑓𝑓 , 𝑧)𝑢2 ] + 𝑁𝑝𝑟𝑜𝑝
2 𝑖
𝑖=1

4.2.3. Flow Angle of Attack Correction to Control Surfaces

The forces and moments on appendages are strongly dependent on the angle of attack, so

proper estimation is important. The higher-order terms for sail and stern planes in the

hydrodynamic loading equations represent the performance change due to angular motions. The

terms can be reduced if the angles of attack to the planes are computed properly, then the

controlling forces and moments from the appendages can be computed without higher-order

corrections. For X-shape stern planes appropriate estimation of flow angle of attack is even more

important, because all stern planes have authority in both horizontal and vertical plane maneuver,

unlike cruciform configuration.

A schematic of angle of attack to the 𝑖th plane is illustrated in Figure 5.

40
Figure 5. The angle of attack correction for the upper portside stern plane

The computation requires a vector representing the axes of plane 𝒃𝑋𝑖 , a vector normal to

𝒃𝑋𝑖 and a vector 𝒑𝑋𝑖 parallel to the cord of the plane without deflection, to form the plane of 𝑖th

appendage in neutral position. Using the surface normal vector 𝒏𝑋𝑖 and the velocity vector at the

center of the appendage plane 𝒖𝑠,𝑋𝑖 in the local coordinate system, the correction angle 𝛼𝑋𝑖 can be

computed as

𝒖𝑠,𝑋𝑖 ∙ 𝒖𝑝,𝑋𝑖
𝛼𝑋𝑖 = cos −1 ( ) (32)
|𝒖𝑠,𝑋𝑖 ||𝒖𝑝,𝑋𝑖 |

𝒖𝑝,𝑋𝑖 = (𝒖𝑠,𝑋𝑖 ∙ 𝒃𝑋𝑖 ) 𝒃𝑋𝑖 + (𝒖𝑠,𝑋𝑖 ∙ 𝒑𝑋𝑖 ) 𝒑𝑋𝑖 (33)

𝒖𝑠𝑖 = 𝒖𝑠 + 𝒓𝑋𝑖 × 𝛀𝑠 (34)

where, the vector 𝒖𝑝,𝑋𝑖 is the projected vector to the plane formed by 𝒃𝑋𝑖 and 𝒑𝑋𝑖 . The velocity

vector to the center of the planes 𝒖𝑠,𝑋𝑖 in still water can be obtained from the angular velocity

vector 𝛀𝑠 and the position vector 𝑒𝑖 to the center of 𝑖th plane.

41
4.2.4. Control Surface Performance in Waves

The velocity field in wave changes the angle of attack to the planes. The velocity vector to

the planes 𝒖𝑠,𝑋𝑖 is corrected by the wave particle velocity vector in body coordinate system 𝒖𝑤𝑠

as

𝒖𝑠,𝑋𝑖 = 𝒖𝑠 − 𝒖𝑤𝑠 + 𝒓𝑋𝑖 × 𝛀𝑠 (35)

4.2.5. Virtual Mass Forces and Moments

A moving submerged body displaces fluid, and the mass of the fluid yields inertia forces

and moments when the body experiences acceleration. The virtual mass is the equivalent mass of

water accelerated by the body, and the force due to acceleration can be given using the virtual mass

coefficient 𝑐𝑣𝑚 as

𝑭𝑣𝑚 = 𝒄𝑣𝑚 𝑚𝒖̇ 𝑠 (36)

The virtual mass is a symmetric 3 × 3 matrix but can be simplified to a diagonal matrix containing

virtual mass coefficient in each direction. The streamlined shape of BB2 makes the virtual mass

force in longitudinal direction reasonably small, while being much larger in lateral and vertical

directions.

The complex geometry of BB2 can be divided into sub-sections, where the virtual mass

coefficient for the 𝑖th section is evaluated independently. The total virtual mass force is then

expressed as the sum of all sectional forces


𝑛
𝑑
𝑭𝑣𝑚 = ∑ 𝒄𝑣𝑚 (𝑖) 𝑚(𝑖) 𝒖 (37)
𝑑𝑡
𝑖=1

where the accuracy of virtual mass force computation is enhanced when more sub sections are

considered. The sectional approach allows computation of the virtual mass moments without

introducing virtual mass coefficients due to angular acceleration.

42
The moment due to virtual mass forces can be computed using the position vector 𝒓𝑖 to the

geometric center of 𝑖th section with respect to the 𝐶𝐺.


𝑛

𝑴𝑣𝑚 = ∑ 𝒓𝑖 × 𝑭𝑣𝑚 (𝑖) (38)


𝑖=1

4.2.6. Virtual Mass Forces and Moments due to Wave Particle Acceleration

The submarine maneuvering in waves is exposed to the wave velocity field. The

acceleration of the wave particles induces additional virtual mass forces and moments which

attenuates with the depth. The total force due to virtual mass can be obtained from the rigid body

and wave velocities 𝑼𝑟,𝑖 and 𝑼𝑤,𝑖 in each section


𝑛
𝑑
𝑭𝑣𝑚 = ∑ 𝒄𝑣𝑚 (𝑖)𝑚(𝑖) (𝑼 (𝑗) + 𝑼𝑟𝑖 (𝑗)) (39)
𝑑𝑡 𝑤 𝑖
𝑖=1

The rigid body velocity 𝑼𝑟,𝑖 of each section is computed from the body angular velocity

vector 𝛀𝑠 and the position vector of the section volumetric center as

𝛀 𝑠 = {𝑝 𝑞 𝑟 }𝑇 (40)

𝑼𝑟,𝑖 = 𝒖𝑠 + 𝛀𝑠 × 𝒓𝑖 (41)

Since equation (38) accounts for the total moment due to virtual mass, the moment due to wave

particle acceleration is computed without modifying or introducing a new equation.

43
4.3. Hydrostatic Forces and Moments

4.3.1. Surface Grid

In order to compute hydrostatic pressure on the surface, a surface geometry is imported

from 14 sub-surfaces containing 7,148 triangular elements as shown in table 5.

Table 5. The list of surface grids used in hydrostatic forces and moments computation

Component Sub-Components Number of Elements


Hull Cylindrical Hull 2,914
Sail Structure 564
Propeller Hub 550
Stern Plane Roots 904
Sail Planes Sail Planes 1 661
Sail Planes 2 661
Stern Planes LowerSt1 200
LowerSt2 200
UpperSt1 200
UpperSt2 200
UpperPt1 200
UpperPt2 200
LowerPt1 200
LowerPt2 200

4.3.2. Computation of Hydrostatic Forces

From the coordinates of three vertices on each triangular element, the hydrostatic pressure

can be evaluated at the mid points of the faces, as illustrated in Figure 6.

Figure 6. Computation of hydrostatic forces on the 𝑖th triangle

44
For 𝑖th triangle, the surface area normal vector 𝑨𝒊 is defined by the product of area of the

triangle 𝐴𝑖 and the unit surface normal vector 𝒏𝒊 as

𝑨𝒊 = 𝐴𝑖 𝒏𝒊 = {𝐴𝑖,𝑥 𝐴𝑖,𝑦 𝐴𝑖,𝑧 }𝑻 (42)

The force vector for 𝑖th triangle 𝑭𝒊 in the 2nd order accuracy is computed from pressures

at the midpoints of each connector and the surface area normal vectors as

1
𝑭𝒊 = − (𝑝12,𝑖 + 𝑝23,𝑖 + 𝑝31,𝑖 )𝑨𝒊 (43)
3

Then the total hydrostatic force 𝑭 obtained by integrating 𝐹𝑖 for all triangular surface elements.

𝑖𝑚𝑎𝑥

𝑭 = ∑ 𝑭𝒊 (44)
𝑖=1

4.3.3. Computation of Hydrostatic Moments

The hydrostatic moment of 𝑖th triangle is obtained from the pressure of midpoint of each

face and corresponding position vector with respect to the 𝐶𝐺, as illustrated in Figure 7.

Figure 7. Computation of hydrostatic moments for the 𝑖th triangle

The hydrostatic moment for the 𝑖th triangle 𝑴𝒊 is computed as

45
1
𝑴𝒊 = − (𝒓𝟏𝟐,𝒊 × 𝑨𝒊 𝑝12,𝑖 + 𝒓𝟐𝟑,𝒊 × 𝑨𝒊 𝑝23,𝑖 + 𝒓𝟑𝟏,𝒊 × 𝑨𝒊 𝑝31,𝑖 )
3
(45)
1
= − (𝒓𝟏𝟐,𝒊 𝑝12,𝑖 + 𝒓𝟐𝟑,𝒊 𝑝23,𝑖 + 𝒓𝟑𝟏,𝒊 𝑝31,𝑖 ) × 𝑨𝒊
3

and the total hydrostatic moment 𝑴 is the result of integrating elemental moment 𝑴𝒊 over all

triangles,

𝑖𝑚𝑎𝑥

𝑴 = ∑ 𝑴𝒊 (46)
𝑖=1

4.3.4. Wave Hydrostatic Pressure

The total hydrostatic pressure under a regular wave at a depth 𝑧 and spatial location 𝑥 can

be expressed as a function of the wave frequency 𝜔𝑤𝑎𝑣𝑒 and wave number 𝑘𝑤𝑎𝑣𝑒 . The variable ℎ

in the equation is the distance from the sea bottom; taking the limit to large ℎ yields an expression

for pressure field at deep water condition.

cosh 𝑘(ℎ + 𝑧)
𝑝 = −𝜌𝑔𝑧 + lim 𝜌𝑔𝐴𝑤𝑎𝑣𝑒 sin(𝑘𝑤𝑎𝑣𝑒 𝑥 − 𝜔𝑤𝑎𝑣𝑒 𝑡)
ℎ→∞ cosh 𝑘ℎ (47)
= −𝜌𝑔𝑧 + 𝜌𝑔𝐴𝑤𝑎𝑣𝑒 𝑒 𝑘𝑧 sin(𝑘𝑤𝑎𝑣𝑒 𝑥 − 𝜔𝑤𝑎𝑣𝑒 𝑡)

4.4. Captive Simulations

Hydrodynamic coefficients are evaluated from a set of captive test simulations.

Simulations are designed to isolate hydrodynamic derivatives appearing in the submarine dynamic

model. Straight line tests are performed to identify propulsion characteristics and hull drift forces

and moments. Rotating arm and pure rotation tests are performed to identify damping

characteristics due to rotational motions. From horizontal and vertical PMM, the transient effects

due to acceleration are evaluated.

46
4.4.1. Self-Propulsion and Straight Towing Simulations

Straight towing simulation and even keel self-propulsion results are used to evaluate the

thrust deduction 𝑡. The thrust 𝑇𝑝𝑟𝑜𝑝 at a given propeller rotational speed 𝑛 is

𝑇𝑝𝑟𝑜𝑝 = (1 − 𝑡) 𝐾𝑇 𝜌𝑛2 𝐷4
(48)
2
𝐾𝑇 = 𝑐2 𝐽 + 𝑐1 𝐽 + 𝑐0

where 𝐾𝑇 is the thrust coefficients, 𝐽 is the propeller advance coefficient defined as 𝐽 = 𝑢/𝑛𝐷, and

𝐷 is diameter of the propeller. The coefficients 𝑐0 , 𝑐1 and 𝑐2 are obtained by fitting the 𝐾𝑇 curve

obtained experimentally or from CFD to a 2nd order polynomial. The thrust is implemented in the

CFD simulations as an actuator disk body force with a Hough and Ordway load distribution (Fu et

al. 2015).

The bare hull resistance is evaluated from the straight ahead towed and self-propulsion

computations for each target speed and depth. For towed simulations the grid does not move in

space, while is only allowed to surge for self-propulsion simulations.

The thrust 𝑇 and resistance 𝑅 in Figure 8 show asymptotic decrease with increasing depth.

The axial forces were non-dimensionlized by 𝐹𝑥 /𝜌𝑈02 𝐿20 for each simulation condition. The

decrease in axial force for 10𝑘𝑡𝑠 between 𝐷0 = 2.5 and 7.0𝑚 is about 15% while the change is

within 3% for 3𝑘𝑡𝑠 and 6𝑘𝑡𝑠. This is due to the larger wave making resistance in approach speed

10𝑘𝑡𝑠, as shown in the flow fields displayed in Figure 9. Notice that the top sail depth 𝐷0 is related

to the center of gravity depth 𝑍𝐶𝐺 by 𝑍𝐶𝐺 = 𝐷0 + 11.37 𝑚.

47
Figure 8. Thrust and resistance as function of approach speed and depth

48
Figure 9. Free surface elevation for self-propelled BB2 at 6𝑘𝑡𝑠 and 10𝑘𝑡𝑠 and two depths.

Thrust deduction is the result of different performance of a propeller in open water and

when attached behind the hull. The required thrust is higher than the bare hull resistance at the

same speed because the propeller attached to the hull accelerates the flow leading to increased

pressure and friction resistance and change in wake distribution into the propeller plane. The thrust

deduction is a concept devised by naval architects to explain such difference,

49
𝑇−𝑅
𝑡𝑑 = (49)
𝑇

The thrust deduction for different depths and approach speeds is shown in Figure 10. The

thrust deduction generally decreases with depth but also decreases as the submarine approaches

close to the free surface, likely due to the free surface pressure boundary condition releasing some

of the propeller-induced pressure loss at the tail.

Figure 10. Thrust deduction for approach speeds 3𝑘𝑡𝑠, 6𝑘𝑡𝑠 and 10𝑘𝑡𝑠.

50
4.4.2. Accelerated Forward Simulations

The virtual mass coefficients are evaluated from 15 subsections of the vehicle, as shown in

the Table 6 and Figure 11. The vehicle is towed with constant acceleration of 1.0𝑚𝑠 −2 in the

longitudinal direction, and 0.1𝑚𝑠 −2 in lateral and vertical directions.

Figure 11. Hull surface sections for virtual mass coefficient evaluation

51
Table 6. Virtual mass coefficients obtained for different sections

Section 𝒄𝒗𝒎𝒙𝒙,𝒊 𝚫𝒎𝒊 𝒄𝒗𝒎𝒚𝒚,𝒊 𝚫𝒎𝒊 𝒄𝒗𝒎𝒛𝒛,𝒊 𝚫𝒎𝒊


Hull 1 8.47E+04 1.72E+05 1.65E+05
Hull 2 2.36E+04 4.34E+05 3.62E+05
Hull 3 1.16E+03 5.98E+05 4.30E+05
Hull 4 1.41E+02 1.14E+06 6.56E+05
Hull 5 9.15E+01 6.97E+05 5.17E+05
Hull 6 9.61E+01 6.72E+05 5.19E+05
Hull 7 1.22E+04 5.78E+05 4.75E+05
Hull 8 3.72E+04 3.47E+05 3.35E+05
Hull 9 1.93E+04 7.73E+04 7.66E+04
Hull 10 1.54E+04 2.64E+04 2.63E+04
Sail 1.83E+04 4.59E+05 5.11E+04
Sail Plane Roots 3.19E+02 1.26E+04 3.49E+02
Stern Plane Roots 3.03E+03 4.49E+04 4.45E+04
Sail Planes 7.92E+02 8.90E+02 1.33E+04
Stern Planes 1.21E+03 3.95E+04 3.92E+04
∑ 𝑐𝑣𝑚𝑗𝑗,𝑖 Δ𝑚𝑖 /𝑚 0.05 1.15 0.81

52
4.4.3. Pure Drift Simulations

Constant speed simulations with vertical and horizontal drift angles α and β were

performed to evaluate hull drift forces and moments. Note that the signs of α and β are determined

such that α is positive when the bottom side of the hull is windward, and β positive when the

starboard side is windward. The range of α and β considered is -12° to 12° with a fixed interval of

4°. Attitude is changed while maintaining the incoming velocity, therefore all hydrostatic

coefficients are obtained as function of total velocity. For each drift angle, the forces and moments

are measured at 𝑈0 = 3, 6 and 10𝑘𝑡𝑠, and 𝐷0 = 2.5, 4, 7 and 25𝑚.

The forces and moments are evaluated in body local coordinate system. The resistance 𝑋 ′

as function of 𝛼 and 𝛽 in Figures 12 and 13 shows that the axial force decreases quickly with

horizontal drift angle, trends already reported by Quick and Woodyatt (2014) for BB1. Simulations

are performed only for positive horizontal drift angles 𝛽 since the hull is symmetric about the

center plane. As a consequence, the coefficients are symmetric for 𝑋′ , 𝑍′ and 𝑀′ , and

antisymmetric for 𝑌 ′ , 𝐾 ′ , and 𝑁 ′ . While the effects of 𝛼 on 𝑌, 𝐾 and 𝑁 are negligible due to the

geometric symmetry, 𝑍 and 𝑀 increase almost linearly with the vertical drift angle. 𝑌 decreases

and 𝐾 and 𝑁 increase linearly proportional to 𝛽. This result implies that the vehicle tends to sway,

roll and yaw to the leeward side, and heave down and pitch up when turning to port in the

horizontal plane. Bow up resistance is higher than bow down.

53
Figure 12. Hull resistance, vertical force and pitch moment as function of depth, speed and 𝛼

54
Figure 13. Hull resistance and lateral and vertical forces and as function of depth, speed and 𝛽

55
Figure 14. Hull roll, pitch and yaw moments as function of depth, speed and 𝛽

56
4.4.4. Straight Towing with Static Control Surface Deflection

Simulations at constant forward speed towing with the control surfaces deflected are used

to evaluate the loads on the planes. The hydrodynamic coefficients for sail and stern planes are

shown in Figures 15 to 23, where 𝐹𝑥,𝛿𝑋𝑖 , 𝐹𝑦,𝛿𝑋𝑖 and 𝐹𝑧,𝛿𝑋𝑖 are the forces and 𝑚𝑥,𝛿𝑋𝑖 , 𝑚𝑦,𝛿𝑋𝑖 and

𝑚𝑧,𝛿𝑋𝑖 are the moments about the 𝐶𝐺. See Eq. (A2) in Appendix A for normalization.

The performance of the control surfaces in general saturate beyond the 𝛿𝑋𝑖 > 10° .

Resistance of all planes increases quadratically with deflection, while other forces and moments

change linearly before stall.

Unlike the stern planes, the resistance of sail planes is different for positive and negative

deflections. Lateral force, roll moment and yaw moment from the sail planes are neglected

according to the Gertler and Hagen (1967) model formulation, since they are symmetric respect to

the center plane and move together.

57
Figure 15. Sail plane resistance, vertical force, and pitch moment as a function of depth, speed and
deflection angle

58
Figure 16. Lower starboard stern plane resistance, lateral force, and vertical force as a function of depth,
speed and deflection angle

59
Figure 17. Lower starboard stern plane roll, pitch, and yaw moments as a function of depth, speed and
deflection angle

60
Figure 18. Upper starboard stern plane resistance, lateral force, and vertical force as a function of depth,
speed and deflection angle

61
Figure 19. Upper starboard stern plane roll, pitch, and yaw moments as a function of depth, speed and
deflection angle

62
Figure 20. Upper portside stern plane resistance, lateral force, and vertical force as a function of depth,
speed and deflection angle

63
Figure 21. Upper portside stern plane roll, pitch, and yaw moments as a function of depth, speed and
deflection angle

64
Figure 22. Lower portside stern plane resistance, lateral force, and vertical force as a function of depth,
speed and deflection angle

65
Figure 23. Lower portside stern plane roll, pitch, and yaw moments as a function of depth, speed and
deflection angle

66
4.4.5. Pure Rotation Simulation

Steady rotation about the center of gravity in roll, pitch and yaw with rates 𝑝, 𝑞 and 𝑟 are

imposed. A schematic of the implementation is shown in the Figure 24.

Figure 24. Pure rotation simulations in vertical 𝑥𝑦 and 𝑦𝑧 planes

The results are direction dependent only for 𝑞, due to geometric symmetry about the 𝑦-

plane. The imposed rotational motion about the center of gravity gives derivatives for 𝑝2 , 𝑞|𝑞| and

𝑟 2 without decoupling. This is not a typical captive submarine model test, because of the

difficulties in implementing the experimental setup. The major components obtained from this test

are damping moment coefficients against the direction of rotation, which are difficult to separate

in RA tests. The linear force coefficients due to rotational motions are also obtained. As discussed

in §2.1, they do not have high authority in plane maneuverability, but the influence in off-plane

maneuver stability is not negligible.

Figures 25 and 26 show the coefficients obtained from the pure rotation tests. See Eq. (A3)

in Appendix A for details of equations used for normalization. Figure 25 shows the coefficients

obtained from the test performed in the xz-plane by imposing constant pitch rate. The in-plane
′ ′ ′
coefficients 𝑋𝑞|𝑞| , 𝑍𝑞|𝑞| and 𝑀𝑞|𝑞| reach an asymptotic with increasing angular velocity, implying

that a quadratic damping behavior is observed for higher pitch rates but linear for smaller ones.

Figure 26 show the coefficients obtained from the pure rotation test in the 𝑦𝑧-plane with

fixed roll rate. The resistance, vertical force, and pitching moment induced by roll motions have

consistent sign regardless of the direction due to geometric symmetry. Trends are similar to those

67
discussed for pitch rate with the exception of the lateral force which is strongly affected by the

geometry of the sail.

The result of constant roll with forward free stream velocity 𝑢𝑝 is shown in Figure 27.The

hull is pushed forward and downward, and pitching downward regardless of the sign of 𝑝 due to

symmetry. The sway force, rolling moment, and yawing moment are dependent on roll rate sign.

Figure 25. Damping forces and moments due to 𝑞

Figure 26. Damping forces (a) and damping moments (b) due to q

68
Figure 27. Damping forces (a) and damping moments (b) due to 𝑢𝑝

4.4.6. Rotating Arm Simulations

RA tests are performed to evaluate the derivatives for terms with coupled translational and

rotational velocities. These coupling velocity components can be isolated by rotating the hull about

a pivot point outside the hull, maintaining a constant flow incidence. Hull, appendages and

refinement grids is rotated about a point from a distance the center of gravity, as described in the

schematic in Figure 28.

Figure 28. Rotating arm simulations in 𝑦𝑧-plane (a) 𝑥𝑧-plane (b) 𝑥𝑦-plane (c)

69
RA tests for each pivot point are conducted for reversed rotation as well, since some terms

are dependent on the direction of rotation. Since terms with coupled translational and rotational

velocities act as major damping sources, the resulting coefficients govern the directional stability

of the hull. The translational velocity in surge, sway or heave is imposed by the leverage distance

𝑅 and the imposed rotational velocity 𝑝, 𝑞 and 𝑟. The motion of the 𝐶𝐺 with constant roll, pitch

and yaw rates 𝑝0 , 𝑞0 , 𝑟0 and leverage length 𝑅0 in the vertical and horizontal RA test is imposed

as

𝑦 = 𝑅0 sin(𝑝0 𝑡)

𝑧 = 𝐷0 + 𝑅0 (1 − cos(𝑝0 𝑡)) (50)

𝜙 = 𝜙0 + 𝑝0 𝑡

𝑥 = 𝑅0 sin(𝑞0 𝑡)

𝑧 = 𝐷0 + 𝑅0 (1 − cos(𝑞0 𝑡)) (51)

𝜃 = 𝜃0 + 𝑞0 𝑡

𝑥 = 𝑅0 sin(𝑟0 𝑡)

𝑦 = 𝑅0 (1 − cos(𝑡0 )) (52)

𝜓 = 𝜓0 + 𝑟0 𝑡

The values 𝜙0 , 𝜃0 and 𝜓0 are zero for 𝑣𝑝, 𝑢𝑞 and 𝑢𝑟 coupled velocities, respectively. The

attitude of the submarine at the initial point changes with these values, therefore 90° for these

values will simulate 𝑤𝑟 , 𝑤𝑞 and 𝑣𝑟 velocities, respectively. The translational velocity of the

submarine is proportional to 𝑅0 and roll/pitch/yaw rate. When RA test is performed in 𝑥𝑧 and 𝑥𝑦


′ ′ ′ ′
planes, an arbitrary static roll angle 𝜙0 can be applied. Since terms 𝑌𝑣|𝑟| , 𝑍𝑤|𝑞| , 𝑀|𝑤|𝑞 and 𝑁|𝑣|𝑟

1
include the influence of the flow angle in the yz-plane, appropriate ranges of (𝑣 2 + 𝑤 2 )2 need to

be simulated.

70
RA tests benefit from CFD because the derivatives in the vertical plane can be estimated

as function of depth. RA tests in 𝑥𝑦-plane do not have restrictions in space, however, in the xz and

yz planes geometric restrictions occur. Since the forces and moments need to converge before the

hull deviate far from the target depth, rotational speeds that are too fast or leverage lengths that are

too large cannot be used. See the Appendix A for the coefficient normalization.

The coefficients obtained from RA tests in the 𝑥𝑧-plane are shown in Figure 29. 𝑋𝑢𝑞 is a


sign dependent component with positive 𝑋𝑢𝑞 indicating additional resistance, and the magnitude

decreases quickly as the magnitude of 𝑢𝑞 increases. As expected the damping force coefficient

due to 𝑢𝑞 motions is higher at smaller 𝑢𝑞 as this implies lower Reynolds numbers, with an

asymptotic coefficient at higher Reynolds number where the force increases with the velocity

squared. In circumstances like a change in direction over a short advance distance and time, the

damping will not be significant. In the current version of the hydrodynamic model, used
′ (−𝑢𝑞) ′ (𝑢𝑞)
𝑋𝑢𝑞 = −𝑋𝑢𝑞 for negative values of 𝑢𝑞.

′ ′ ′
𝑍𝑢𝑞 and 𝑀𝑢𝑞 show similar behavior to 𝑋𝑢𝑞 . The results show that a positive 𝑢𝑞 induces

′ ′
downward heave force and bow down pitch moment. 𝑋𝑤𝑞 , 𝑍𝑤𝑞 are obtained from the set where

the vertical drift angle is 90°. Since the terms are sign dependent in the model, opposite sign is

applied for the negative range of 𝑤𝑞. The evaluation for those coefficients should be performed,

since certain level of difference exists due to the asymmetry of the geometry. For positive 𝑞 (which

also results in positive 𝑤𝑞), the results show that the submarine is pushed forward and downward,

and a damping moment to 𝑀𝐻 is generated.


′ ′ ′
For values of 𝑢𝑞 and 𝑤𝑞 beyond the largest simulated, 𝑋𝑢𝑞 , 𝑍𝑢𝑞 and 𝑀𝑢𝑞 are linearly

′ ′ ′
extrapolated, and 𝑋𝑤𝑞 , 𝑍𝑤𝑞 and 𝑀𝑤𝑞 are assumed to be constant with same value of the highest

71
′ ′ ′
evaluated point. For 𝑋𝑤𝑞 , 𝑍𝑤𝑞 and 𝑀𝑤𝑞 the convergence of the trend is not clear, therefore

evaluation for larger 𝑤𝑞 is required to improve the model quality.

Figure 30 shows the coefficients for coupled surge/roll. 𝑢𝑟 induces a small level of

propulsion (positive 𝑋𝑢|𝑟| ), downward force and sway force and pitching moment opposing the

sign of roll.

Figure 29. Hydrodynamic forces (a) and moments (b) obtained from RA tests in 𝑥𝑧-plane

Figure 30. Hydrodynamic forces (a) and moments (b) obtained from RA tests in 𝑥𝑦-plane

72
4.4.7. Propeller side forces and moments

A single screw propeller generates side forces and moments as well as thrust. In addition,

the propeller in rotational body motions generate counteracting forces and moments. They are

proportional to 𝑛 and the angular rates 𝑝 or 𝑟 depending on the DoF of interest. The loading terms
′ ′ ′
(𝑛, 𝑢𝑟) , 𝑍𝑝𝑟𝑜𝑝 ′ ′
(𝑛, 𝑝), 𝑀𝑝𝑟𝑜𝑝 ′
𝑋𝑝𝑟𝑜𝑝 (𝑛, 𝑣𝑟, 𝑤𝑞) , 𝑌𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑞) , 𝐾𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑞) and 𝑁𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑟) are

modeled using data from free running vertical and horizontal zigzag CFD simulations performed
′ ′ ′
with discretized overset propeller. 𝑋𝑝𝑟𝑜𝑝 (𝑛, 𝑤𝑞), 𝑍𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑞), and 𝑀𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑞) are modeled

′ ′ (𝑛, 𝑢𝑟), and


from vertical zig zag simulations (Carrica et al. 2021), and 𝑋𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑟) , 𝑌𝑝𝑟𝑜𝑝


𝑁𝑝𝑟𝑜𝑝 (𝑛, 𝑢𝑟) are modeled from horizontal zig zag simulation results (Carrica et al. 2020). For

details of coefficient normalization, see Eq. (A6) in the appendix A.

The propeller side forces and moments, and the fluctuation of thrust and resistance are

shown in Figures 31 (a) and (b) respectively. The magnitude normalized forces and moments are

decreasing as the approach speed decreases.

The propeller rolling moment 𝐾𝑝𝑟𝑜𝑝 is in principle dependent on the submarine motions.

However, motions of the vessel do not have significant influence in the propeller rolling moment,

since propeller torque is weakly affected by the motions. In this sense, propeller rolling moment

coefficient 𝐾𝑛′ is assumed to be only dependent on 𝑛 and the approach speed as shown in Figure

32.

73
Figure 31. Propeller side forces and moments (a) variation of thrust and resistance (b) as function
of speed

Figure 32. Propeller roll moment as function of speed and 𝑛

74
5. VALIDATION OF HYDRODYNAMIC MODEL

5.1. Free Roll Decay Test

A submarine rotated around its longitudinal axis is subject to a restoring moment due to

the combined effect of gravity and buoyancy. This moment increases linearly with the magnitude

of angular excursion and the distance between 𝐶𝐺 and 𝐵𝐺, and vanishes when the two centers are

vertically aligned. Due to the submarine geometry, the restoring moment due to roll is much

smaller than the restoring pitching moment and a slow decay of the roll can be measured through

an experiment for which the vessel is released from a static angular displacement and let roll until

the periodic motion decays by friction and drag. Under this conditions, the peak amplitude decays

exponentially, while the oscillation period remains constant. This experiment, and other described

in this chapter, are used to validate the hydrodynamic model by isolating responses of the model

to specific inputs.

HDM results for a case with an initial roll angle 𝜙0 = 4.58°, are shown in Figure 33

compared against the experimental data of Overpelt et al. (2015). The roll decay prediction follows

closely the experiment result for the initial 3~4 periods, then the error accumulates. The period of

motion is also slightly overpredicted by HDM. Note that the HDM simulation is performed in full-

scale while the experiment is at model scale, which may have caused the damping to be

underpredicted.

75
Figure 33. Free roll decay test against model experiment

5.2. Vertical Zig Zag Maneuver (VZZ)

A controlled vertical zigzag (VZZ) maneuver is considered next. In this maneuver, starting

from self-propulsion, the control planes are actuated at their maximum speed to a set angle and

maintained at this position until the submarine reaches a check pitch angle, at which time the

control surfaces are actuated in the opposite direction. This process is then repeated over two

cycles. For validation of the model, CFD results for 3, 6, 10, and 15 knots at full scale are used.

The sail plane is only used at 3 and 6𝑘𝑡𝑠 (the deflection is set to zero at high speed). The

deflection of each plane is determined by Eq. (1).

The approach speed is obtained with a speed controller to achieve self-propulsion

conditions with controller setpoints at zero pitch, yaw, sway and depth change. The propeller

rotational speed 𝑛 is then fixed to the self-propulsion value during the zigzag maneuver. Velocity

decreases during the maneuver due to increased resistance from the deflected appendages and the

operation at a drift angle.

76
In all VZZs considered herein, 𝛿𝑉 is set to ±10° with maximum rudder rate of 7.11°/s

and the check point 𝜃 = ±10°. Since the inertia of body is very large compared to the forces and

moments produced by control planes after they reverse at the check point, the pitch angle continues

increasing, resulting in a few degrees overshoot angle in each direction.

In this validation test, time histories of pitch angle, pitch rate, velocity decrease, and

vertical excursion are compared against CFD simulation results. The time series are presented as

function of dimensionless time normalized by the time elapsed until the first positive overshoot 𝑇1

of CFD simulations.

The pitch angle is shown in Figure 34. Overshoot angles and the period of motion are in

good agreement between the two methods, with the largest errors among them observed at low

speed. Pitch rates in Figure 35 are also in good agreement with CFD. The forward velocity, and

the vertical excursion, shown in Figs. 36 and 37, are also overall in good agreement. The

differences between the two methods do not present clear trends with speed that would allow a

simple re-calibration of the model. Notice for instance that the error in vertical displacement at the

first overshoot is largest at 6 knots, however the velocity reduction is also smallest in CFD at this

speed, and not well aligned with other CFD results, which makes it difficult to further draw

conclusions from these results.

The appendage vertical forces and pitching moments for 𝑈0 = 10 kts are shown in Figure

38. The appendage loading in the initial stage of maneuver is very important, since the hull drift

angle is initially small and contributes little to the maneuvering forces and moments. The

comparison in the figure suggests that the proposed computation of 𝛿𝑋𝑒𝑓𝑓,𝑋𝑖 in §4.2.3 predicts the

controlling force with good accuracy without the need of high-order terms as used in the

conventional formulation. Both vertical force and pitching moment are slightly overpredicted by

77
HDM; notice that as already discussed, added mass is not accounted for appendage deflection.

Considering that the added mass always opposes the motion, the difference is not significant.

Consequently, the initial slope of 𝜃 and 𝑞 in Figure 34 and 35 show good agreement against CFD.

Figure 34. Time series of 𝜃 during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts.

Figure 35. Time series of 𝑞 during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts.

78
Figure 36. Time series of velocity during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts.

Figure 37. Time series of depth change during VZZ 10/10 at 3kts, 6kts, 10kts and 15kts.

79
Figure 38. Time series of appendage total vertical force (𝑎) and total pitching moment (b) from
sail and stern planes at 10kts

5.3. Horizontal Zig Zag Maneuver (HZZ)

A horizontal zigzag (HZZ) maneuver is executed with the same approach as a VZZ

maneuver, using the horizontal command to control yaw, instead of pitch. The maneuver is started

after controlled self-propulsion is achieved at the desired approach speed. The horizontal

command 𝛿𝐻 controls the stern planes, starting the maneuver either to port or starboard at

maximum rudder rate. As the submarine yaw angle 𝜓 increases until it reaches its check point in

each direction, the stern planes are executed in the opposite direction, producing a horizontal

zigzag trajectory. Additional control is required to maintain depth, which is achieve actuating on

the stern planes via the vertical command 𝛿𝑉 .

A single case of HZZ (𝛿𝐻 = ±20°, checking angle 10° and speed 𝑈0 = 10 kts) simulated

using HDM is compared with experimental and CFD results in Figure 39. In the initial stage, the

yawing moment of the stern planes determines the rate of yaw, in the same fashion as the

appendage pitching moment dominates the pitch rate in VZZ. The slope of the curves in 𝜓 and 𝑟

show good agreement with CFD and EFD, resulting in good agreement of the first overshoot angle.

80
Speed decrease is the key factor of the maneuver, since boat speed governs all

hydrodynamic loadings. The overall decrease is consistent with CFD and EFD, indicating that the

influence of thrust and resistance variation in rotational motion is well captured. Roll angle is also

in good agreement against benchmark data while the roll rate is slightly underpredicted by HDM.

Figure 39. The 20/20 HZZ motions in model test, CFD simulation and HDM simulation at 10kts

81
5.4. Self-Propulsion in Waves

An important requirement of the HDM is that it can predict correctly the operation of the

submarine near the surface. To test its ability dynamically, self-propulsion in waves is considered

next. The submarine moves freely during the maneuver, being affected by the virtual mass due to

wave particle acceleration and wave hydrostatic forces and moments. Therefore, the amplitude of

heave and pitch are good indicators of wave loads. For validation, the first harmonics of heave and

pitch in a 6-DoF controlled self-propulsion are compared against CFD results at 𝑈0 = 10 kts and

𝐷0 = 23.69, 4 and 2.5 m in regular waves equivalent to sea state 5. The simulations used both sail

and stern planes and ballast tank to control depth. For details of CFD set up and controller gains,

see Carrica et al. (2019).

The first harmonic depth change and pitch amplitude for HDM are approximately

83~95% and 87~94% of the CFD result, respectively. Note that the vertical motion in CFD is

approximately 27% the amplitude of the wave particle motion, indicating that the controller is not

able to handle the vertical displacement of the vehicle in large sea states. The results show that the

wave hydrostatic and added mass forces and moments computation in the dynamic model provide

a good representation of the wave effect on the vehicle.

Figure 40. First harmonic of depth change normalized by wave amplitude (a) and 𝜃 (b) from self-
propulsion in regular waves SS5

82
5.5. Horizontal Turning Circle Maneuver

In this maneuver, the horizontal command 𝛿𝐻 is fixed to a target value for the submarine

to perform a turning circle, while the horizontal plane is maintained by controlling depth via the

vertical command 𝛿𝑉 .

A set of benchmark cases is chosen for validation from CFD simulations performed for

𝐷0 = 23.69, 7, 4, and 2.5 m at 𝑈0 = 10 kts in calm water and in regular waves equivalent to a

sea state 5. The speed control is active to maintain the approach speed, and 𝛿𝐻 is fixed to 20° to

turn to starboard. Both sail and stern planes are used and the ballast tank is active to assist with

depth control.

The trajectories of all turning circles in calm water are compared in Figure 41. The radius

of the circle is about 200 m for all cases, with HDM producing slightly larger circle in all

conditions. The radius decreases with depth for both methods. The yaw rate 𝑟 in the initial 90°of

the maneuver is in good agreement against CFD indicating that the yawing moment from the stern

planes is well predicted by HDM.

83
Figure 41. Trajectory during a turning circle maneuver in calm water at 10kts

Figure 42 shows the trajectory in regular waves. Drift is expected as the vessel repeats the

circle due to changing wave direction and CFD results show this drift in operation near the surface,

while the trajectories for HDM show very little drift. This illustrate a limitation of the HDM for

which further adjustments might be needed to properly predict drift in waves.

84
Figure 42. Trajectory during turning circle maneuver in regular waves SS5 at 10kts

Figures 43 to 45 show the evolution of depth change and pitch angle with heading for all

depths considered. Overall trends for pitch angle coincide for both methods. In following seas

conditions (heading angle of 180°), HDM exhibits the largest fluctuation in pitch, while the

magnitude is less than half of CFD. The differences for vertical displacement 𝑑𝑧 are also largest

in following seas, as the vertical controller is unable to maintain the target depth in those

conditions. The environmental load is underestimated in HDM for all test conditions, due to the

lack of an implemented mechanism to address wave drift forces.

85
The environmental load is underestimated in HDM for all test conditions, due to the lack

of an implemented mechanism to address wave drift forces.

Figure 43. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at the sail
top depth23.69 m as function of heading angle

Figure 44. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at the sail
top depth 4 m as function of heading angle

86
Figure 45. Pitch angle (a) and depth change (b) for a 10 kts turning circle maneuver at sail top
depth 2.5 m as function of heading angle

5.6. Max-q Maneuver

This benchmark test assesses the ability of a submarine to perform emergency rise. The

vertical command 𝛿𝑉 is imposed to the desired value during the maneuver, while the vertical plane

is maintained by the horizontal controller. The sail planes are not used during this maneuver. The

maximum pitch angle 𝜃𝑚𝑎𝑥 , and the time history of pitch angle and pitch rate are the variables of

major interest.

Results from EFD and CFD in model scale for an emergency rise at equivalent speed of

𝑈0 = 12 kts, fixing the vertical command to 𝛿𝑉 = 20°m are used for the evaluation of HDM. The

maneuver is initiated from 𝑧𝐶𝐺 = −40𝑚, and the propeller rotational speed is fixed at the self-

propulsion value. The comparison is only valid until surfacing, as HDM is designed for underwater

operation only. The nose of the submarine reaches the surface in the experiment at around 25 s

and earlier for HDM, due to a smaller reduction in speed that causes a faster emergence of the

submarine (Figure 46). Overall, pitch angle and pitch rate in Figure 46 (a) and (c) show excellent

87
match in the initial stage of the maneuver, until about 10 seconds, which implies that the stern

plane moment is predicted well; both the maximum pitch rate and when it occurs are also well

captured. Error accumulates beyond the peak of pitch rate, accumulating an error of about 5° in

pitch when the vehicle finishes surfacing at 𝑡~32𝑠 . The strong deviation in forward speed

indicates that the model might not perform as accurately in extreme conditions which exceed those

used for parameter calibration.

Figure 46. The time series of motions during max-q maenuver: (a) 𝜃 (b) speed decrease (c) pitch
rate (d) vertical travel distance

88
6. SUMMARY, CONCLUSIONS, LIMITATIONS AND FUTURE WORK

6.1. Summary

In this thesis a hydrodynamic model of the generic submarine Joubert BB2 operating near

the surface and in waves is presented. The standard maneuvering equation by Gertler and Hagen

(1967) was revised for submarines with X-shape stern planes and a pair of sail planes attached to

the sail structure. Individual hydrodynamic coefficients for a generic design submarine Joubert

BB2 are obtained from towing, pure rotation, and rotating arm simulations. Propeller thrust is

modeled with the Hough and Ordway actuator disc model using POW results for the MARIN

7371R propeller, and the thrust deduction evaluated from straight towing and even-keel self-

propulsion simulations. Changes in thrust, side forces and moments of the propeller due to angular

motions of the submarine are modeled from available VZZ and HZZ CFD simulation data. The

maneuvering equations and hydrodynamic coefficients are implemented into MATLABTM

SIMULINKTM to simulate submarine motions by solving rigid body equations of motion in calm

water and regular waves.

The control force estimation during maneuvers, conventionally done using higher order

terms, is simplified to a new form that uses coefficients in even-keel condition and effective flow

angle of attack. Added mass is evaluated from several subsections of the submarine, to allow

prediction of virtual mass effects due to wave particle acceleration. Hydrostatic loads due to ship

attitude change and wave phase are computed by integrating pressure on surface elements of a

triangulated Joubert BB2 geometry.

PID controllers are implemented to simulate controlled maneuvers. The algorithm consists

of independent vertical and horizontal controllers acting on the sail and stern planes, speed

controller acting on the propeller rotational speed, 𝐿𝐶𝐺 controller acting on the trim tanks, and

89
mass controller operating the ballast tank. The implemented controllers are used to perform

validation and benchmark studies in §5.

6.2. Conclusions

The high-order terms appearing in the original Gertler and Hagen (1967) model can be

truncated by substituting the appendage deflection angle 𝛿𝑋𝑋𝑖 by 𝛿𝑋𝑒𝑓𝑓,𝑋𝑖 . From the effective flow

angle of attack to the appendages computed by §4.2.3 a single set of coefficients from even-keel

condition in §4.4.4 yields accurate performance estimation.

Added mass and hydrostatic load computations in §4.2.3 and §4.2.3 yield reasonable

prediction for wave loads, as discussed in §5.4. Computing virtual mass from several sections

allows calculation of the virtual mass moments without additional set of coefficients for angular

acceleration. Since the added mass always opposes the motion it dampens the overall motions

response. The results for VZZ and HZZ in §5.2 and §5.3 show that the damping from added mass

was reasonably estimated as the submarine experiences extremely unsteady motions. The

hydrostatic load also predicts the restoring moment of the submarine, which limits the maximum

pitch and pitch rate that the submarine can reach. The results of VZZ simulation, and the max-q

maneuver in §5.6 indicate that the restoring moment is computed properly.

The Hough and Ordway actuator disc model predicts reasonable propeller thrust in

motions, if thrust variation with incidence angle is included in the model as shown in §4.4.7. CFD

results with discretized propeller can be used for modeling propeller side forces and moments. The

decrease in thrust due to motions governs the speed decrease in VZZ and HZZ at fixed 𝑛. Since

the advance speed determines the control forces from appendages, excellent agreement in VZZ

motions and good agreement in HZZ means that the implementation of propeller thrust model

appears to be adequate.

90
6.3. Limitations

There are four major limitations in the model.

First, the wave drift velocity is not considered. This does not bring about serious problems

in head wave conditions, however, results in underpredicted response in beam and following waves

as shown in §5.4. Additionally, the absence of the wave drift means that the wave only has

oscillatory loads, thus the shift of turning trajectory is not predicted.

Second, the effect of wave particle velocity and accelerations to the propeller performance

is not accounted for. This may not have large influence when maneuvering at depth but it becomes

more significant in near the surface as wave particle motions become larger.

Third, the virtual mass computation does not account for the effect of control surface

deflection. The appendage added mass changes as they deviate from the neutral position, however,

HDM computes the virtual mass assuming the neutral deflection. This leads to errors in damping

forces when the control surfaces are actuated, resulting in accumulated errors in motion response.

Lastly, the effect of the free surface is only partially implemented in the model. The

interaction between the hull and the free surface at depths smaller than 𝐷0 = 2.5𝑚 has not been

included, and the model is also not capable of simulating the condition where the submarine is

partially emerged. Since the most extreme maneuvering conditions obviously include the

possibility of accidental emergence (Kim et al. 2020), the inability to simulate such conditions is

a limitation.

91
6.4. Future Work

Short-term and long-term improvements are planned for the HDM.

Improvements to account for wave drift loads and partially emerged condition are planned

in the short term. The solution to the problem of wave drift forces observed in §5.4 requires

modeling od additional load terms. The terms can be modeled as drag forces arising from wave

velocity as conceptually suggested in Fischer et al. (2014) and Wang et al. (2015). By evaluating

drag coefficients for sub sections as done in §4.4.2, the model will be capable of computing wave

drag forces and moments. The partial emergence of the body leads to loss of buoyancy and change

in center of action of the buoyancy force. Since buoyancy is computed from hydrostatic loads in

§4.3.2 and §4.3.3, modifying related subroutines in the program to account for the position of the

free surface blanking cells completely or partially emerged would suffice.

Estimation of propeller performance in waves can be improved by adding wave-induced

particle velocities at the propeller plane. This can be implemented relatively easily, by adding the

local wave velocity when computing the advance coefficient.

The influence of control surface deflection to the added mass will require additional set of

hydrodynamic coefficients, obtained from accelerated towing simulations with several plane

deflection angles, as well as accelerations on the deflection of the planes. The procedure entails

significant amount of computational resources, but the procedures would be similar to those used

to obtain other coefficients.

92
REFERENCES

Behara, S., Arnold, A., Martin, J.E., Harwood, C.M., Carrica, P.M., “Experimental and
computational study of operation of an amphibious craft in calm water,” Ocean Engineering Vol.
209, Article 107460, 2020.

Bettle, M., “Unsteady Computational Fluid Dynamics Simulations of Six Degrees-of Freedom
Submarine Manoeuvres,” Doctoral Thesis, The University of Brunswick, 2013.

Broglia R., Di Mascio A., Muscari R., “Numerical study of confined water effects on a self-
propelled submarine in steady manoeuvres”, The Sixteenth International Offshore and Polar
Engineering Conference, International Society of Offshore and Polar Engineers, San Francisco,
2006.

Carrica, P.M., Wilson, R.V., Noack, R.W., Stern, F., “Ship motions using single-phase level set
with dynamic overset grids,” Computers and Fluids, Vol. 36, pp. 1415-1433, 2007a.

Carrica, P.M., Wilson, R.V., Stern, F., An unsteady single-phase level set method for viscous free
surface flows. International Journal for Numerical Methods in Fluids, Vol. 53, pp. 229-256, 2007b.

Carrica, P.M., Paik, K., Hosseini, H., Stern, F., “URANS Analysis of a Broaching Event in
Irregular Quartering Seas,” Journal of Marine Science and Technology Vol. 13, pp. 395-407, 2008

Carrica, P.M., Huang, J., Noack, R., Kaushik, D., Smith, B., Stern, F., “Large-Scale DES
Computations of the Forward Speed Diffraction and Pitch and Heave Problems for a Surface
Combatant,” Computers and Fluids Vol. 39, pp. 1095-1111, 2010.

Carrica, P.M., Kerkvliet, M., Quadvlieg, F., Pontarelli, M., Martin, E., “CFD Simulations and
Experiments of a Maneuvering Generic Submarine and Prognosis for Simulation of Near Surface
Operation,” 31st Symposium on Naval Hydrodynamics, Monterey, California, 2016.

Carrica, P.M., Kim, Y., Martin, J.E., “Near-Surface Operation of a Generic Submarine in Calm
Water and Waves,” Ocean Engineering Vol. 183, pp. 87-105, 2019.

Carrica, P.M., Kerkvliet, M., Quadvlieg, F., Matin, J.E., “CFD Simulations and Experiments of a
Submarine in Turn, Zigzag, and Surfacing Maneuvers,” Journal of Ship Research, Vol. 64, pp. 1-
16, 2020.

Carrica, P.M., Kim, Y., Matin, J.E., “Vertical zigzag maneuver of a generic submarine,” Ocean
Engineering Vol. 219, Article 108386, 2021.

Chase, N., Michael, T., Carrica, P.M., “Overset simulations of a submarine in towed, self-
propelled and maneuvering conditions,” Internaltional Shipbuilding Progress, Vol. 60, pp. 171-
205, 2013.

93
Cho, Y.J, Seok, W., Cheon, K.H., Rhee, S.H., “Maneuvering simulation of an X-plane submarine
using computational fluid dynamics,” International Journal of Naval Architecture and Ocean
Engineering Vol. 12, pp. 843-855, 2020.

Crook, T.P., “An initial assessment of free surface effects on submerged bodies,” Master Thesis,
Naval Postgraduate School, Monterey, CA, 1994.

Dawson, E., “An Investigation into the effects of submergence depth, speed and hull length-to-
diameter ratio on the near surface operation of conventional submarines,” Ph.D. Thesis, University
of Tasmania, 2014.

Evans, J., Nahon, M., “Dynamics modeling and performance evaluation of an autonomous
underwater vehicle,” Ocean Engineering Vol. 31, pp. 1835-1858, 2004.

Faltinsen, O.M., Minsaas, K.J. and Liapis, N.S., SO 1980. Prediction of Resistance and Propulsion
of a Ship in a Seaway. In Proc. 13th Symposium on Naval Hydrodynamics, Tokyo.

Fang, M.C., Chang, P.E., Luo, J.H., “Wave effects on ascending and descending motions of the
autonomous underwater vehicle,” Ocean Engineering Vol. 33, pp. 1972-1999, 2006.

Feldman, J., “Straightline and Rotating Arm Captive-model Experiments to Investigate the
stability and Control Characteristics of submarine and Other Submerged Vehicles,” David Taylor
Research Center, Ship Hydromechanics Department, Bethesda MD, 1987.

Feldman, J., “Method of Performing Captive-model Experiments to Predict the Stability and
Control Characteristics of Submarines,” Carderock Division Naval Surface Warfare Center
CRDKNSWC-HD-0393-25, 1995.

Fischer, N., Hughes, D., Walters, P., Schwartz, E.M., Dixon, W.E., “Nonlinear RISE-Based
Control of an Autonomous Underwater Vehicle,” IEEE Transactions on Robotics Vol.30, No.4,
2014.

Fu, H.P., Michael, T.J., Carrica, P.M., “Computation on self-propulsion at ship point based on a
body-force propeller,” Journal of Ship Mechanics Vol. 19 (7), pp. 791-796, 2015

Gao, T., Wang, Y., Pang, Y., Chen, Q., Tang, Y., “A time-efficient CFD approach for
hydrodynamic coefficient determination and model simplification of submarine,” Ocean
Engineering Vol. 154, pp. 16-26, 2018.

Geisbert, J.S., “Hydrodynamic modeling for autonomous underwater vehicles using computational
and semi-empirical methods,” Master Thesis, Virginia Polytechnic Institute and State University,
Blacksburg, VA, 2007

Gertler, M., “Resistance Experiments on a Systematic Series of Streamlined Bodies of Revolution-


Epplication to the Design of High-Speed Submarines,” DTIC Report C-297, April 1950.

Gertler, M. and Hagen G.R., “Standard Equations of Motion for Submarine Simulation,” DTIC
Document, Tech. Rep., 1967.

94
Go, G., Ahn, H.T., “Hydrodynamic derivative determination based on CFD and motion simulation
for a tow-fish,” Applied Ocean Research, Volume 82, January, pp. 191-209, 2019.

Goodman, A., “Experimental Techniques and Methods of Analysis used in Submerged Body
Research,” 3rd Symposium on Naval Hydrodynamics, Washington D.C., 1960.

Gourlay, T., Dawson, E., “A Havelock Source Panel Method for Near-surface Submarines,”
Journal of Marine Science and Application Vol. 14, pp. 215-224, 2015.

Hough, G.R., Ordway, D.E., “The generalized actuator disk,” Developments in Theoretical and
Applied Mechanics, Vol. 2, pp. 317-336, 1965.

Huang, J., Carrica, P.M., Stern, F., “Semi-coupled air/water immersed boundary approach for
curvilinear dynamic overset grids with application to ship hydrodynamics,” International Journal
for Numerical Methods in Fluids Vol. 58, pp. 591-624, 2008.

Isa, K., Arshad, M.R., Ishak, S., “A hybrid-driven underwater glider model, hydrodynamics
estimation, and an analysis of the motion control,” Ocean Engineering Vol. 81, pp. 111-129, 2014.

ITTC, “Recommended Procedures and guidelines: Practical Guidelines for Ship CFD
Applications,” International Towing Tank conference, 7.5-03-02-03., 2011

Joubert, P.N., “Some aspects of submarine design part 1,” Australia Defence Science and
Technology Organisation Report DSTO-TR-1622, 2004.

Joubert, P.N., “Some aspects of submarine design part 2,” Australia Defence Science and
Technology Organisation Report DSTO-TR-1920, 2006.

Kerwin, J.E., Kinnas, S.A., Lee, J.-T., Shih, W.-Z., “A Surface Panel Method for the
Hydrodynamic Analysis of Ducted Propellers,” Transaction fo Society of Naval Architecture and
Marine Engineering Vol. 95, pp. 93-122, 1987.

Kim, C.H., Chou, F.S., Tien, D., “Motions and hydrodynamic loads of a ship advancing in oblique
waves,” SNAME Transactions Vol. 88, pp. 225-256, 1980.

Kim, H., Ranmuthugala, D., Leong, Z.Q., Chin, C., “Six-DOF simulations of an underwater
vehicle undergoing straight line and steady turning manoevers,” Ocean Engineering Vol.150,
pp.102-112, 2018.

Kim, Y., Martin, J.E., Carrica, P.M., “Vertical Zigzag and Controlled Turn Maneuvers of a Generic
Submarine in Deep Water and Near the Surface in Irregular Waves,”, 33rd Symposium on Naval
Hydrodynamics, Osaka, Japan, 2020.

Kim, Y., Martin, J.E., Rober, N., Cichella, V., Carrica, P.M., “A dynamic maneuvering model for
the Joubert BB2 submarine in calm water and waves”, Ocean Engineering, to be published, 2021.

95
Li, J., Yuan B., Carrica, P.M., “Modeling Bubble Entrainment and Transport for Ship wakes:
Progress using Hybrid RANS/LES Methods,” Journal of Ship Research, Vol. 64, Article 09180071
2020.

Li, Y., Castro, A.M., Martin, J.E., Sinokrot, T., Prescott, W., Carrica P.M., “Coupled
computational fluid dynamics/multibody dynamics method for wind turbine aero-servo-elastic
simulation including drivetrain dynamics,” Renewable energy, Vol. 101, pp. 1037-1051, 2017.

Li, Y., Martin, J.E., Michael, T., Carrica, P.M., “A Study of Propeller Operation Near a Free
Surface,” Journal of Ship Research, Vol. 59, pp. 1-11, 2015.

Martin, J.E., Michael, T.J., Carrica, P.M., “Submarine Maneuvers Using Direct Overset
Simulation of Appendages and Propeller and Coupled CFD/Potential Flow Propeller Solver,”
Journal of Ship Research Vol. 59, No. 1, pp. 31-48, 2015.

Mansoorzadeh, Sh., Javanmard, E., “An investigation of free surface effects on drag and lift
coefficients of an autonomous underwater vehicle (AUV) using computational and experimental
fluid dynamics methods,” Journal of Fluids and Structures Vol. 51, pp. 161-171, 2014.

Menter, F., “Two Equation Eddy-Viscosity Turbulence Modeling for Engineering Applications.
AIAA Journal Vol. 32, pp. 1598-1605, 1994.

Mohseni, K., “Pulsatile vortex generators for low-speed maneuvering of small underwater
vehicles,” Ocean Engineering Vol. 33, pp. 2209-2223, 2006.

Nematollahi, A., Dadvand, A., Dawoodian, M., “An axisymmetric underwater vehicle-free surface
interaction: A numerical study”, Ocean Engineering Vol. 96, pp. 205-214, 2015.

Noack, R., Boger, D., Kunz, R., Carrica, P.M., “Suggar++: An Improved General Overset Grid
Assembly Capability,” 19th AIAA Computational Fluid Dynamics Conference, AIAA Paper
2009-3992, San Antonio, TX, 2009.

Overpelt, B., Nienhuis, B., Anderson, B., “Free Running Manoeuvring Model Tests On A Modern
Generic SSK Class Submarine (BB2),” Pacific 2015, Sydney, Australia, 2015.

Pan, Y., Zhang, H., Zhou, Q., “Numerical Prediction of Submarine Hydrodynamic Coefficients
Using CFD Simulation,” Journal of Hydrodynamics, Vol. 24, pp. 840-847, 2012.

Pan, Y., Zhang, H., Zhou, Q., “Numerical simulation of unsteady propeller force for a submarine
in straight ahead sailing and steady diving maneuver,” International Journal of Naval Architecture
and Ocean Engineering 11, 2019.

Papoulias, F.A., Riedel, J.S., “Solution Branching and Dive Plane Reversal of Submarines at Low
Speeds,” Journal of Ship Research Vol. 38, pp. 203-212, 1994.

Paik, K., Carrica, P.M., Lee, D., Maki, K., “Strongly Coupled Fluid-Structure Interaction Method
for Structural Loads on Surface Ships,” Ocean Engineering, Vol. 36, pp. 1346-1357, 2009.

96
Park, J.Y., Kim, N., Shin, Y.K., “Experimental Study on Hydrodynamic coefficients for high-
incidence-angle maneuver of a submarine,” International Journal of Naval Architecture and Ocean
Engineering Vol. 9, Issue 1, January 2017, pp. 100-113, 2017.

Phillips, A.B., Turnock, S.R., Furlong, M., “The use of computational fluid dynamics to aid cost-
effective hydrodynamic design of autonomous underwater vehicles,” Engineering for the Maritime
Environment Vol. 224, pp. 239-254, 2010.

Quick, H., Woodyatt, B., “Phase II Experimental Testing of a Generic Submarine Model in the
DSTO Low Speed Wind Tunnel,” DSTO report, Australia, August 2014.

Rober, N., Cichella, V., Martin, J.E., Kim, Y., Carrica, P.M., “3D Path-Following Control for an
underwater Vehicle”, Journal of Guidance, to be published, 2021.

Rober, N., Cichella, V., Martin, J.E., Kim, Y., Carrica, P.M., “L1 Adaptive Control for Underwater
Vehicles”, To be published, 2021.

Sakamoto, N., Carrica, P.M., Stern, F., “URANS simulations of static and dynamic maneuvering
for surface combatant: part 2. Analysis and validation for local flow characteristics,” Journal of
Marine Scienc and Technology Vol. 17, 446-468, 2012.

Shariati, S.K., Mousaviazadegan, S.H., “The effect of appendages on the hydrodynamic


characteristics of an underwater vehicle near the free surface,” Applied Ocean Research Vol.67
pp. 31-43, 2017.

Teo, K., An, E., Beaujean, P-P.J., “A Robust Fuzzy Autonomous Underwater Vehicle (AUV)
Docking Approach for Unknown Current Disturbances,” IEEE Journal of Ocean Engineering Vol.
37, Issue 2, 2012.

Tyagi, A., Sen, D., “Calculation of transverse hydrodynamic coefficients using computational fluid
dynamic approach,” Ocean Engineering Vol.33, pp. 798-809, 2006.

Wang, C., Zhang, F., Schaefer, D., “Dynamic modeling of an autonomous underwater vehicle,”
Journal of Marine Science and Technology Vol.20 pp. 199-212, 2014.

Wang, L.L., Wang, H.J., Pan, L.X., Guo, J.X., “Active disturbance rejection fuzzy controller for
roll stabilization of autonomous underwater vehicle under wave disturbance,” Discrete Dynamics
in Nature and Society Vol. 2015, Article 835126, 2015.

Watt, G.D., “Modelling and Simulating Unsteady Six Degrees-of-freedom Submarine Rising
Maneuvers,” DRDC Atlantic TR 2007-008, 2007.

Watt, G.D., Bohlmann, H.J., “Submarine rising stability: quasi-steady theory and unsteady
effects,” 25th Symposium on Naval Hydrodynamics, St. John's, 2004.

Tian, W., Song, B., Ding, H., “Numerical research on the influence of surface waves on the
hydrodynamic performance of an AUV,” Ocean Engineering Vol. 183, pp. 40-56, 2019.

97
Xing, T., Shao, J., Stern, F., “BKW-RS-DES of unsteady vortical flow for KVLCC2 at large drift
angles,” Proceedings of 9th International Conference on Numerical Ship Hydrodynamics, pp. 5-8,
Ann Arbor, MI, 2007.

Xing, T., Carrica, P.M., Stern, F., “Large-Scale RANS and DDES Computations of KVLCC2 at
Drift Angle 0 Degree,” CFD Workshop Gothenburg 2010, Gothenburg, Sweden, 2010.

Zhang, H., Xu, Y., Cai, H., “Using CFD Software to Calculate Hydrodynamic Coefficients,”
Journal of Marine Science and Application, Vol.9, pp.149-155, 2010.

98
APPENDIX. Hydrodynamic Coefficients Normalization

The hydrodynamic coefficients appearing in the dynamic model equations in §3 are

obtained from normalization using reference variables, where 𝛼 is the vertical drift angle, 𝛽 is the

horizontal drift angle, 𝜌 is density of the fluid, 𝑈0 is the reference velocity of the CFD simulation

case and 𝐿0 is the reference length of the submarine. The total velocity is defined as 𝑈 =

√𝑢2 + 𝑣 2 + 𝑤 2 .

Hull drift forces and moments

The appendage forces and moments are excluded when normalizing the loads.

𝑋 ′ (𝑈, 𝛼, 𝛽, 𝑧) = 𝐹𝑥 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿20

𝑌 ′ (𝑈, 𝛼, 𝛽, 𝑧) = 𝐹𝑦 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿20

𝑍 ′ (𝑈, 𝛼, 𝛽, 𝑧) = 𝐹𝑧 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿20


(A1)
𝐾 ′ (𝑈,
𝛼, 𝛽, 𝑧) = 𝑚𝑥 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿30

𝑀′ (𝑈, 𝛼, 𝛽, 𝑧) = 𝑚𝑦 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿30

𝑁 ′ (𝑈, 𝛼, 𝛽, 𝑧) = 𝑚𝑧 (𝑈, 𝛼, 𝛽, 𝑧)⁄2𝜌𝑈02 𝐿30

Control Surface Forces and Moments



𝑋𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝐹𝑥,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿20


𝑌𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝐹𝑦,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿20


𝑍𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝐹𝑧,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿20
(A2)

𝐾𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝑚𝑥,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿30


𝑀𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝑚𝑦,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿30


𝑁𝛿𝑋 𝑖
(𝑈, 𝛿𝑋𝑖 , 𝑧) = 𝑚𝑧,𝛿𝑋𝑖 (𝑈, 𝛿𝑋𝑖 , 𝑧)⁄2𝜌𝑈02 𝐿30

99
Pure Rotation Simulations
′ (𝑞,
𝑋𝑞𝑞 𝑧) = 2𝐹𝑥 (𝑞, 𝑧)⁄𝜌𝑞 2 𝐿40

′ (𝑞,
𝑌𝑞𝑞 𝑧) = 2𝐹𝑦 (𝑞, 𝑧)⁄𝜌𝑞 2 𝐿40

′ (𝑞,
𝑍𝑞𝑞 𝑧) = 2𝐹𝑧 (𝑞, 𝑧)⁄𝜌𝑞 2 𝐿40


𝑀𝑞|𝑞| = 2𝑚𝑦 (𝑞, 𝑧)/𝜌𝑞|𝑞|𝐿50

′ (𝑟,
𝑋𝑟𝑟 𝑧) = 2𝐹𝑥 (𝑟, 𝑧)⁄𝜌𝑟 2 𝐿40
′ (𝑟,
𝑌𝑟𝑟 𝑧) = 2𝐹𝑦 (𝑟, 𝑧)⁄𝜌𝑟 2 𝐿40

′ (𝑟,
𝑍𝑟𝑟 𝑧) = 2𝐹𝑧 (𝑟, 𝑧)⁄𝜌𝑟 2 𝐿40 (A3)

𝑁𝑟|𝑟| = 2𝑚𝑧 (𝑟, 𝑧)/𝜌𝑟|𝑟|𝐿50


𝑌𝑝|𝑝| = 2𝑚𝑧 (𝑝, 𝑧)/𝜌𝑝2 𝐿40


𝑍𝑝𝑝 = 2𝑚𝑧 (𝑝, 𝑧)/𝜌𝑝2 𝐿40


𝐾𝑝𝑝 = 2𝑚𝑥 (𝑝, 𝑧)/𝜌𝑝2 𝐿50


𝑀𝑝𝑝 = 2𝑚𝑦 (𝑝, 𝑧)/𝜌𝑝2 𝐿50


𝑀𝑟𝑟 = 2𝑚𝑦 (𝑟, 𝑧)/𝜌𝑟 2 𝐿50

100
Rotating Arm Simulations

To isolate the terms as function of the product of translational velocity and angular

velocity, results from towing and pure rotation tests are required. All coefficients are evaluated

without control surface loadings.

𝑣 𝜋
′ (𝑤,
𝑋𝑤𝑞 𝑞, 𝑧) = 2𝐹𝑥 (𝑤, 𝑞, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑋𝑢𝑣𝑤

(𝑈, ′ (𝑞,
, 0, 𝑧) /𝐿0 − 𝑋𝑞𝑞 𝑧)𝐿0
|𝑣| 2

𝑣 𝜋
′ (𝑣,
𝑋𝑣𝑟 𝑟, 𝑧) = 2𝐹𝑥 (𝑣, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑋𝑢𝑣𝑤

(𝑈, 0, ′ (𝑟,
, 𝑧) /𝐿0 − 𝑋𝑟𝑟 𝑧)𝐿0
|𝑣| 2

𝑣 𝜋

𝑌𝑤𝑝 𝑤𝑝 = 2𝐹𝑥 (𝑣, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑌𝑢𝑣𝑤

(𝑈, 0, ′ (𝑟,
, 𝑧) /𝐿0 − 𝑌𝑟𝑟 𝑧)𝐿0
|𝑣| 2

𝑌𝑟′ = 2𝐹𝑦 (𝑈, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑌𝑢𝑣𝑤


′ (𝑈, ′ (𝑟,
0,0, 𝑧)/𝐿0 − 𝑌𝑟𝑟 𝑧)𝐿0
(A4)
𝑣 𝜋
′ (𝑣,
𝑍𝑣𝑟 𝑟, 𝑧) = 2 𝐹𝑧 (𝑣, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑍𝑢𝑣𝑤

(𝑈, 0, ′ (𝑟,
, 𝑧)⁄𝐿0 − 𝑍𝑟𝑟 𝑧)𝐿0
|𝑣| 2

𝑍𝑞′ (𝑢, 𝑞, 𝑧) = 2𝐹𝑧 (𝑢, 𝑞, 𝑧)⁄2𝜌𝑈02 𝐿30 − 𝑍𝑢𝑣𝑤


′ ′ (𝑞,
(𝑈, 0,0, 𝑧) − 𝑍𝑞𝑞 𝑧)

′ 𝑤 𝜋
𝑍𝑤|𝑞| (𝑢, 𝛽, 𝑞, 𝑧) = 2 𝐹𝑧 (𝑢, 𝛽, 𝑞, 𝑧)⁄𝜌𝑈02 𝐿30 − 𝑍𝑢𝑣𝑤

(𝑈, °, 𝛽, 𝑧)⁄𝐿0
|𝑤| 2
′ (𝑟,
−𝑍𝑞𝑞 𝑧)𝐿0

𝐾𝑟′ = 2𝑚𝑦 (𝑢, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑀𝑢𝑣𝑤


′ ′ (𝑟,
(𝑈, 0,0, 𝑧)/𝐿0 − 𝑀𝑟𝑟 𝑧)𝐿0

𝑤 𝜋

𝐾𝑤𝑝 = 2𝑚𝑦 (𝑤, 𝑝, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑋𝑢𝑣𝑤

(𝑈, ′ (𝑟,
, 0, 𝑧) /𝐿0 − 𝑀𝑝𝑝 𝑧)𝐿0
|𝑤| 2

𝑣 𝜋

𝑀𝑣𝑟 = 2 𝑚𝑦 (𝑣, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑀𝑢𝑣𝑤

(𝑈, 0, ′ (𝑟,
, 𝑧)⁄𝐿0 − 𝑀𝑟𝑟 𝑧)𝐿0
|𝑣| 2

𝑀𝑞′ = 2𝑚𝑦 (𝑈, 𝑞, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑀𝑢𝑣𝑤


′ ′ (𝑟,
(𝑈, 0,0, 𝑧)/𝐿0 − 𝑀𝑞𝑞 𝑧)𝐿0

′ 𝑣 𝜋
𝑀|𝑤|𝑞 = 2 𝑚𝑦 (𝑤, 𝛽, 𝑞, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑀𝑢𝑣𝑤

(𝑈, ′ (𝑟,
, 𝛽, 𝑧)⁄𝐿0 − 𝑀𝑞𝑞 𝑧)𝐿0
|𝑣| 2

𝑣 𝜋

𝑁𝑤𝑝 = 2𝑚𝑧 (𝑤, 𝑝, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑁𝑢𝑣𝑤

(𝑈, ′ (𝑟,
, 0, 𝑧) /𝐿0 − 𝑁𝑝𝑝 𝑧)𝐿0
|𝑣| 2

101
𝑁𝑟′ = 2𝑚𝑧 (𝑢, 𝑟, 𝑧)⁄𝜌𝑈02 𝐿40 − 𝑁𝑢𝑣𝑤
′ ′ (𝑟,
(𝑈, 0,0, 𝑧)/𝐿0 − 𝑁𝑟𝑟 𝑧)𝐿0

102
Forced Acceleration

The hydrodynamic coefficients for acceleration can be evaluated from accelerating the

submarine with constant rate. The unsteady terms can be obtained from vertical or horizontal PMM

tests. To isolate the unsteady terms, the results from pure rotation test and towing test. When the

loadings are measured using PMM test, the rotating arm test results are also required.

𝑋𝑢̇′ = 2𝐹𝑥 ⁄𝜌𝐿30 𝑢̇ − 𝑋𝑢𝑣𝑤


′ (𝑈, 0,0, 𝑧)𝐿0 𝑈 2 /𝑢̇

𝑌𝑟̇′ = 2𝐹𝑦 /𝜌𝐿40 𝑟̇


𝑌𝑝̇′ 𝑝̇ = 2𝐹𝑦 ⁄𝜌𝐿40 𝑝̇ − 𝑌𝑝|𝑝| 𝑝|𝑝|⁄𝑝̇

𝑣 𝜋
𝑌𝑣̇′ = 2𝐹𝑦 ⁄𝜌𝐿30 𝑣̇ − 𝑌𝑢𝑣𝑤

(𝑈, 0, , 𝑧) 𝑈 2 /𝐿0 𝑣̇
|𝑣| 2

𝑍𝑞̇′ = 2𝐹𝑧 /𝜌𝐿40 𝑞̇ (A5)



𝑀𝑞̇′ = 2𝑚𝑦 ⁄𝜌𝐿50 𝑞̇ − 𝑀𝑞|𝑞| 𝑞|𝑞|/𝑞̇

𝑣 𝜋
𝐾𝑣̇′ = 2𝑚𝑥 ⁄𝜌𝐿40 𝑣̇ − 𝐾𝑢𝑣𝑤

(𝑈, 0, , 𝑧) 𝑈 2 /𝐿0 𝑣̇
|𝑣| 2

𝑁𝑟̇′ = 2𝑚𝑧 ⁄𝜌𝐿50 𝑟̇ − 𝑁𝑟|𝑟| 𝑟|𝑟|/𝑟̇

𝑁𝑝̇′ = 2𝑚𝑧 ⁄𝜌𝐿50 𝑝̇

Propeller Side Forces and Moments

The hydrodynamic coefficients for acceleration can be evaluated from accelerating the

submarine with constant rate. The unsteady terms can be obtained from vertical or horizontal PMM

tests. To isolate the unsteady terms, results from pure rotation and towing tests are used. When the

loadings are measured using PMM tests, rotating arm test results are also required.

𝑋𝑝𝑟𝑜𝑝 = 𝐹𝑥,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿40 (𝑢𝑞 + 𝑢𝑟) (A6)

103

𝑌𝑝𝑟𝑜𝑝 = 𝐹𝑦,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿40 𝑢𝑟


𝑍𝑝𝑟𝑜𝑝 = 𝐹𝑧,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿40 𝑢𝑞


𝐾𝑝𝑟𝑜𝑝 = 𝑚𝑥,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿50 𝑛2


𝑀𝑝𝑟𝑜𝑝 = 𝑚𝑦,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿50 𝑢𝑞


𝑁𝑝𝑟𝑜𝑝 = 𝑚𝑧,𝑝𝑟𝑜𝑝 /2𝜌𝑈02 𝐿50 𝑢𝑟

104

You might also like