+ All Categories
Home > Documents > NUMERICAL SIMULATION TOOL FOR MOORED …13068/datastream... · NUMERICAL SIMULATION TOOL FOR MOORED...

NUMERICAL SIMULATION TOOL FOR MOORED …13068/datastream... · NUMERICAL SIMULATION TOOL FOR MOORED...

Date post: 03-Jul-2018
Category:
Upload: vukhue
View: 238 times
Download: 0 times
Share this document with a friend
78
NUMERICAL SIMULATION TOOL FOR MOORED MARINE HYDROKINETIC TURBINES by Basil L. Hacker Jr. A Thesis Submitted to the Faculty of The College of Engineering and Computer Science in Partial Fulfillment of the Requirements for the Degree of Master of Science Florida Atlantic University Boca Raton, Florida December 2013
Transcript

NUMERICAL SIMULATION TOOL FOR MOORED MARINE HYDROKINETIC

TURBINES

by

Basil L. Hacker Jr.

A Thesis Submitted to the Faculty of

The College of Engineering and Computer Science

in Partial Fulfillment of the Requirements for the Degree of

Master of Science

Florida Atlantic University

Boca Raton, Florida

December 2013

iii

ACKNOWLEDGEMENTS

My greatest thanks go to my thesis advisor Dr. James VanZwieten because without his

support and guidance of this project would not have been possible for me.

I would also like to thank Dr. Palaniswamy Ananthakrishnan for his valuable contributions in

the role as my co-advisor and to Dr. Manhar Dhanak and Dr. Gopal Gaonkar as my committee

members.

I would like to thank and dedicate my thesis to my loving and supportive family, with special

mention for mention for my mom and little brother who have been a major driving force in all that I

have accomplished. This family includes but is not limited to blood as I have had many valuable

friends who have also played a major role in supporting me in my studies who I also consider to be

family.

I would also like to thank the people at FAU’s Southeast National Marine Renewable Energy

Center for taking me on as a Master’s student, allowing me to work on an interesting project, whose

funding that helped to support me and this research and finally made me feel at home while working

with them.

iv

ABSTRACT

Author:

Basil L. Hacker Jr.

Title:

Numerical Simulation Tool for Moored Marine Hydrokinetic Turbines

Institution: Florida Atlantic University

Thesis Co-Advisors:

Dr. Palaniswamy Ananthakrishnan,

Dr. James Van Zwieten Jr.

Degree:

Master of Science

Year:

2013

The research presented in this thesis utilizes Blade Element Momentum (BEM) theory with a

dynamic wake model to customize the OrcaFlex numeric simulation platform in order to allow

modeling of moored Ocean Current Turbines (OCTs). This work merges the advanced cable modeling

tools available within OrcaFlex with well documented BEM rotor modeling approach creating a

combined tool that was not previously available for predicting the performance of moored ocean

current turbines. This tool allows ocean current turbine developers to predict and optimize the

performance of their devices and mooring systems before deploying these systems at sea. The BEM

rotor model was written in C++ to create a back-end tool that is fed continuously updated data on the

OCT’s orientation and velocities as the simulation is running. The custom designed code was written

specifically so that it could operate within the OrcaFlex environment. An approach for numerically

modeling the entire OCT system is presented, which accounts for the additional degree of freedom

(rotor rotational velocity) that is not accounted for in the OrcaFlex equations of motion. The properties

of the numerically modeled OCT were then set to match those of a previously numerically modeled

Southeast National Marine Renewable Energy Center (SNMREC) OCT system and comparisons were

made. Evaluated conditions include: uniform axial and off axis currents, as well as axial and off axis

v

wave fields. For comparison purposes these conditions were applied to a geodetically fixed rotor,

showing nearly identical results for the steady conditions but varied, in most cases still acceptable

accuracy, for the wave environment. Finally, this entire moored OCT system was evaluated in a

dynamic environment to help quantify the expected behavioral response of SNMREC’s turbine under

uniform current.

vi

NUMERICAL SIMULATION TOOL FOR MOORED MARINE HYDROKINETIC

TURBINES

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

LIST OF TABLES ................................................................................................................................. ix

1 INTRODUCTION AND OBJECTIVES ....................................................................................... 1

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

1.2 State of Technology ................................................................................................... 3

1.2.1 Device Design ...................................................................................................... 3

1.2.2 Simulation Developments .................................................................................... 4

1.3 SNMREC’s Experimental OCT ................................................................................. 5

1.4 Contributions and Outline .......................................................................................... 7

2 LITERATURE REVIEW............................................................................................................... 9

2.1 Rotor Modeling .......................................................................................................... 9

2.1.1 Basic Blade Element Momentum Model ............................................................. 9

2.1.2 Complete Blade Element Momentum Model ..................................................... 12

2.1.3 Unsteady Blade Element Momentum Rotor Model ........................................... 16

2.2 OrcaFlex ................................................................................................................... 18

2.2.1 OrcaFlex Coordinate Systems ............................................................................ 19

2.2.2 Elements ............................................................................................................. 20

2.2.3 External Functions .............................................................................................. 22

2.2.4 Equations of motion ........................................................................................... 23

2.2.5 External Buoy Forces ......................................................................................... 25

3 OCT MODEL DEVELOPMENT ................................................................................................ 29

3.1 Coordinate Systems and Kinematics ........................................................................ 29

3.1.1 Transformation Matrices .................................................................................... 31

3.2 Equations of Motion................................................................................................. 32

3.3 Hydrodynamic Modeling ......................................................................................... 35

3.3.1 Rotor Modeling .................................................................................................. 35

3.3.2 Streamlined Body Forces ................................................................................... 43

3.4 Properties of SNMREC’s OCT ................................................................................ 44

3.4.1 Mass, Displace Volume and Inertial Properties ................................................. 44

vii

3.4.2 Hydrodynamic Properties ................................................................................... 45

4 NUMERICAL RESULT .............................................................................................................. 50

4.1 Model Validation ..................................................................................................... 50

4.1.1 Steady State ........................................................................................................ 50

4.1.2 Transient State .................................................................................................... 54

4.1.3 Waves ................................................................................................................. 57

4.2 Moored Turbine Performance .................................................................................. 62

5 CONCLUSIONS AND SUGGESTED FUTURE WORK .......................................................... 65

5.1 Conclusions .............................................................................................................. 65

5.2 Suggested Future Work ............................................................................................ 67

REFERENCES ...................................................................................................................................... 68

viii

LIST OF FIGURES

Figure 1: HYCOM calculated mean kinetic energy flux for 2009-2011 at a depth of 50 m. The

location and value of the maximum temporally averaged hydrokinetic energy flux are

indicated in each plot ................................................................................................................. 3 Figure 2: Artist rendition of SNMREC’s Experimental Ocean Current Turbine .................................... 6 Figure 3: Artist rendering of SNMREC’s turbine mooring system and turbine. Please note that

this figure is not drawn to scale ................................................................................................. 7 Figure 4: (a) Schematic of blade elements; c airfoil chord length, dr radial length of each element,

r is radius, R is rotor radius andΩ is the angular velocity of the rotor. (b) Schematic of the

blade elements with respect to the rotor plane area. ................................................................ 10 Figure 5: Local loads on a blade ........................................................................................................... 11 Figure 6: Control Volume shaped as an annular element used with the BEM model ........................... 13 Figure 7: Image of OrcaFlex’s Coordinate systems .............................................................................. 20 Figure 8: Screen Shot of OrcaFlex’s Graphic User Interface. ............................................................... 21 Figure 9: Lumped buoy coordinate system. .......................................................................................... 30 Figure 10: 3D Lift and Drag Coefficient as angle a function of attack for all possible angles of

attack ....................................................................................................................................... 47 Figure 11: 3D Lift and Drag Coefficients as a function of angle of attack over the expected

operating conditions ................................................................................................................ 47 Figure 12: The predicted coefficients of power, thrust (drag) and torque ............................................. 52 Figure 13: Predicted coefficient of power (top left), power (top right) and restoring moment

comparison from a 3 m rotor diameter operating at a TSR of 4.10. ........................................ 54 Figure 14: Predicted power response when RPM is changed from 41.76 (TSR 4.10) to 20.37 (TSR

2.0) ........................................................................................................................................... 55 Figure 15: Predicted power response when flow speed is changed from 1.6 m/s to 1.4 m/s ................ 56 Figure 16: Shaft Power for a constant RPM of 41.76 experiencing waves ........................................... 58 Figure 17: Forces experienced by a rotor shaft with a fixed RPM of 41.76 when subjected to

waves ....................................................................................................................................... 59 Figure 18: Shaft moments experienced by a rotor with a fixed RPM of 41.76 when subjected to

waves ....................................................................................................................................... 59 Figure 19: Shaft power for a constant RPM of 41.76 operating at a depth of 20 m with incoming

waves that are 90° off axis. ...................................................................................................... 60 Figure 20: Forces applied to the rotor for a constant RPM of 41.76 operating at a depth of 20 m

with incoming waves that are 90° off axis............................................................................... 61 Figure 21: Moments experienced by the rotor for a constant RPM of 41.76 operating at a depth of

20 m with incoming waves that are 90° off axis. .................................................................... 61 Figure 22: Mooring configuration utilized to simulate OCT performance in a uniform steady

current ...................................................................................................................................... 62 Figure 23: This figure shows the power versus time for the rotor for a steady 1.6 m/s current ............ 62 Figure 24:This figure shows the rotation of the OCT about the x-axis (roll) for a steady 1.6 m/s

current with a constant rotor rotational velocity 41.76 RPM................................................... 63 Figure 25: This figure shows the tension in the mooring line attached to the OCT in relation to

time for steady 1.6m/s current with a constant rotor rotational velocity 41.76 RPM. ............. 63

ix

LIST OF TABLES

Table 1: Eight top HYCOM estimated regions in the world capable of providing hydrokinetic

current energy on a large scale, along with their respective areas above three power

intensity thresholds. ................................................................................................................... 2 Table 2: Some of SNMREC’s Ocean Current Turbines predicted specifications ................................... 7 Table 3: System mass, centers of Gravity, and mass moments with respect to the center of mass of

the rotor and the center of buoyancy that are plugged in to OrcaFlex(Products of inertia

for x-y and y-z plane are zero due to symmetry of OCT). ....................................................... 45 Table 4: Inertial characteristics of each component of the OCT system. .............................................. 45 Table 5: This table shows the values for , and at a flow speed of 1.6 m⁄s calculated

using the developed rotor model. ............................................................................................ 53 Table 6: Comparisons of standard deviations and variances of the turbine code when subjected to

axial waves and 90° off axis waves ......................................................................................... 60

1

1 INTRODUCTION AND OBJECTIVES

1.1 Motivation

In order to cope with the increasing demand of energy, people have begun looking to the sea

where they found a large, clean renewable source that avoids many of the environmental issues which

arise from our current major methods of producing electricity. The oceans embody a vast amount of

heat and mechanical energy, which continually moves an enormous volume of water [1]. Within the

vast ocean there are a multitude of sources from which electricity can be extracted; such as ocean

thermal gradients, offshore wind and ocean kinetic energy in the forms of waves and currents. It has

been estimated that the ocean stores enough energy in the form of heat, currents, waves, and tides to

meet total worldwide demand for power many times over [2]. Hydrokinetic and wind power seem to

be the optimal choices in renewable energy sources of today due in large part to the number of sites

that they can be utilized [3], though another potential contender for marine renewable energy

production in the tropics and sub-tropics is ocean thermal energy.

Focusing on the hydrokinetic energy available within the ocean, there are essentially two

means of generating electricity from marine and tidal currents: a) building a tidal barrage across high

tide areas, some including estuaries or bays and b) extracting energy from free-flowing water in the

open ocean [4]. The first mentioned extraction methodology harnesses the potential energy but

presents several issues due their high construction costs and extreme environmental impact [5].Due to

the issues related to this method, alternative means were developed to harness the kinetic energy in

marine currents by placing Marine Hydro-Kinetic Turbines (MHKTs) to absorb the energy in the

water’s velocity without significantly impeding the flow [6].These devices can harness hydrokinetic

energy from river current, tidal current and open ocean currents, the latter of which is the focus of this

thesis.

2

Throughout the world there are several currents with the potential to generate significant

amounts energy. The National Ocean Partnership Program sponsored a multi-institutional consortium

as a means to develop the HYbrid Coordinate Ocean Model (HYCOM) which provides, among other

things, real-time and archived fine resolution ocean current and thermal estimates [7]. Global

HYCOM current data taken daily over a 3 year period, from January 1 2009 – December 31 2011, was

used the calculate the kinetic energy flux density and then averaged at a depth of 50 meters to predict

the ocean current resource globally [8]. That data was then compiled and analyzed resulting in eight

different regions around the globe that contain significantly large areas of electricity production

potential that are presented in Table 1 [8].

Table 1: Eight top HYCOM estimated regions in the world capable of providing hydrokinetic current energy on

a large scale, along with their respective areas above three power intensity thresholds [8].

There are three currents with the largest areas where the power intensity is over 1 kW/m2 are

the Gulf Stream (Figure 1.a) Kuroshio Current (Figure 1.b) and Agulhas Current (South) (Figure

1.c).One of the main drawbacks of the Kuroshio Current (Figure 1.b) off of the coast of Japan is that

much of the energy is not available for Japan to harvest. Due to the treaty known as the Laws of the

Sea of 1982 in which a country is allowed only 200 nautical miles off of their coastline as an exclusive

economic zone, which at latitude of 36° N would only allow Japan access to 4.11 degrees longitude

3

from their coast to harness. The Gulf Stream (Figure 1.a) and the Agulhas Current (South) (Figure 1.c)

are both form fitting to landmasses which allows for more energy to be harnessed and more easily be

fed onto the electric grid.

Figure 1: HYCOM calculated mean kinetic energy flux for 2009-2011 at a depth of 50 m. The location and value

of the maximum temporally averaged hydrokinetic energy flux are indicated in each plot [8]

The Gulf Stream has some unique characteristics due in large part to its passage between the

Bahamas and Southeast Florida through the Florida Straits, which minimizes its opportunity for cross

track meandering. The Gulf Stream also has the largest predicted power intensity at the projected

operating depth of 50 meters below the mean sea level [9]. The maximum average power intensity in

the Gulf Stream at this depth is predicted to be 1.93 kW/m2, compared to 1.78 kW/m

2 for the Kuroshio

Current and 1.66 kW/m2 for the Agulhas Current [8].

1.2 State of Technology

1.2.1 Device Design

Hydrokinetic energy’s toe-hold in the renewable energy market spawns from its versatility of

installation sites, from rivers and ocean currents to human-made channels [10]. MHKTs extract energy

from marine currents under much the same principles as wind turbines and the basic design types for

these systems fall into similar classifications. Most MHKTs are typically classified into two groups,

either axial flow or cross flow turbines [11]. Axial flow turbines are designed to operate with the rotor

axis aligned with the current direction and the system must either be actively or passively controlled to

follow the current direction. Conversely, if the rotational axis is designed to operate perpendicular to

4

the current the flow direction is not important and this design type is referred to as a cross flow turbine

[12]. Axial flow turbines currently have achieved a dominant position in the turbine market, with most

open ocean designs utilizing this rotor type [2, 13, 14, 15]. Most MHKT systems are designed to be

either bottom mounted or moored, but our focus will be on the moored designs since this is the only

feasible method for most open ocean locations.

As with most emerging technologies MHKTs still have several obstacles to overcome before

becoming a fully viable asset in harnessing the energy of the ocean currents. One of the largest hurdles

is designing a system that anchors the turbine to the ground while allowing it to operate near the

surface and align itself with the current. Another hurdle lies in the deployment, operation, and

maintenance and retrieval stage of a MHKT system in the open ocean.

1.2.2 Simulation Developments

Due to some of the hurdles mentioned above the MHKT industry’s is making slow but steady

progress towards deploying systems in the open ocean. Without means to easily test and simulate

designs it is difficult to predict what designs stand the best chance at working, avoiding failures, and

creating electricity that is economically sustainable. An important step towards getting MHKTs

operating in the open ocean is to create a tool that developers can use to simulate the performance of

their coupled turbine and mooring designs.

When modeling large complex systems such as a moored MHKT designed to operate in the

open ocean, it is important to study both the forces on the rotor and the coupled affects that the rotor

and environment have on the moored system as a whole. To model these coupled interactions, rotor

and turbine systems can be modeled and simulated with time domain approaches that account for

system motion when calculating the fluid loadings on the structure. Using this approach, researchers

have gained an initial understanding of the behavior of OCT systems when operating in an offshore

environment [9, 16]. This approach allows all the parts of a complete system to work together without

limiting the ability of each component to be analyzed separately. In order to accurately simulate the

relationship between a rotor and the flow of a fluid body through the swept rotor plane area in a

5

dynamic environment it is important to know both the undisturbed flow at the current time step and

the flow reduction caused by the rotor on the flow from the wake of the previous time steps [17]. This

adds complexity to creating numeric models since numeric modeling programs such as OrcaFlex

assume that the flow is undisturbed by the structure for all of its standard calculations.

Recently, two numerical simulation models have be developed, the first of which utilizes a

BEM rotor model and is implemented in MATLAB/Simulink[18] and the second of which utilizes a

Blade Element (BE) rotor model and is implemented in OrcaFlex [19,20]. The MATLAB model

utilizes several different corrections/modifications to the traditional BEM model that allow it to

account for shear currents and waves. One drawback to this numeric model is that it utilizes a

somewhat basic cable model that limits its usefulness for analyzing commercial OCT systems, which

utilize complex mooring system. A second drawback to this numeric model is that it does not have a

user interface and requires both basic programming skills and experience with numeric simulations to

create meaningful results. The OrcaFlex model, by only incorporating the basic BE rotor model, has

significant inaccuracies in rotor performance calculations and was designed more to analyze the

behavior of the mooring system than the turbine itself. By combining the accurate rotor model

implemented in [18] with the mooring simulation model [19, 20] into a single tool with the best

features of both models, OCT developers will be better equipped to evaluate prototyped systems.

1.3 SNMREC’s Experimental OCT

The Southeast National Marine Renewable Energy Center (SNMREC) has developed an

experiment 3.08 m rotor diameter Ocean Current Turbine (OCT) that is designed to work in open

ocean conditions and generate approximately 20 kW of power in 2.3m/s while operating at a tip speed

around 5.0 [9]. The preliminary design can be seen in Figure 2.

6

Figure 2: Artist rendition of SNMREC’s Experimental Ocean Current Turbine

This experimental OCT is modeled following SNMREC’s intended single point mooring

system that consists of an onsite mooring buoy as well as a support vessel or platform to be towed out

to the test site with the experimental turbine onboard and connected to the mooring buoy after the final

dry tests have been completed (Figure 3)[20]. The experimental turbine will be launched and retrieved

using the A-frame on the support vessel and all monitoring and control applications will be conducted

from a dock van located on this vessel [9]. Since this experimental OCT will likely produce the first

publically available in situ data set of OCT performance that can be utilized for validating numerical

models it is the system that will be numerically modeled in this thesis.

7

Figure 3: Artist rendering of SNMREC’s turbine mooring system and turbine. Please note that this figure is not

drawn to scale [9]

The current prototype design is a negatively buoyant horizontal axis turbine with a rotor

diameter of approximately 3 meters. The performance predictions for this rotor are listed in Table 2

[9]. An important distinction to make is the difference between rated power and average shaft power,

for all practical purposes the rated power is the manufactures maximum power the device can output

without failing and generally requires certain conditions to achieve while the average shaft power is

the amount of power, that due to the intended environmental conditions, a developer can expect the

turbine to regularly produce.

1.4 Contributions and Outline

This thesis combines the methodologies developed in the two previous numeric models and

dynamical simulation [18, 20] in order to create a more accurate modeling tool for simulating moored

Table 2: Some of SNMREC’s Ocean Current Turbines predicted specifications [9.]

8

OCTs. This process consists of modifying the existing MATLAB unsteady BEM model presented in

[9, 18] and re-writing it in a C derivative programming language. This C file is used to create a

dynamic link library file which allows the model to interact with the OrcaFlex buoy, vessel and cable

modeling software as well as native windows applications. Validation of this numeric rotor model is

performed for steady state, transient and steady state current with wave excitation operating

conditions.

By integrating this rotor model into the OrcaFlex environment and utilizing and updated version

of the mooring system presented in [19, 20] for FAU’s SNMREC experimental OCT, the coupled

effects on the OCT system can be calculated. Simulations run using this model are used to predict the

performance of this device and these results are compared with MATLAB predictions for similar

operating conditions as part of the validation process. These performance data are quantified

according to the methodologies suggested by [21] so that they can be compared to future offshore

testing data.

This thesis is organized in four logically progressing chapters after this introduction. Chapter 2

presents a literature review that covers rotor modeling. It also introduces the mooring simulation

package, OrcaFlex, and how the rotor model is interfaces with the OrcaFlex simulation environment.

Next, Chapter 3 provides detailed description of the equations used in this simulation, as well as the

assumptions that were made when creating this tool. This leads to the validation of the rotor code as

well as estimated performance predictions of SNMREC’s OCT, which are presented in Chapter 4.

Finally, Chapter 5 highlights important findings and draws conclusions from this work.

9

2 LITERATURE REVIEW

This chapter covers two fundamental areas that this thesis builds upon. First, several rotor

modeling approaches are discussed, with a focus on the Blade Element Momentum rotor modeling

technique utilized in this thesis. This discussion introduces the fundamental Blade Element Model and

Momentum Model and then discusses modifications that that have been developed so that these

fundamental theories can be used in an operating environment where the fluid and rotor velocities

vary both in time and space. Following this discussion some of the relevant characteristics of the

numeric modeling platform that is utilized in this thesis, OrcaFlex, are discussed. This includes a

general overview of how this modeling platform operates followed by more discussion of OrcaFlex’s

“external function”, a function that allows the user to incorporate custom numerical models into its

operating environment.

2.1 Rotor Modeling

When numerically predicting rotor performance there are many techniques that can be used

including but not limited to: Generalized Dynamic Wake Model (GDW) [22] to Strip Theory or Blade

Element Momentum (BEM) model [23] and Vortex Lattice Method [24,25] just to name a few. The

method that is utilized in this application will be a modified version of the Blade Element Momentum

method and will build upon the work of [18, 9]. The BEM model, in its most basic form, is designed

to model rotor behavior in a steady state [17, 26]. With a few adaptations that will be explained in

Subsections 2.1.2 and 2.1.3 this approach can be modified to allow the BEM model to work for

unsteady states and also increase its accuracy [17].

2.1.1 Basic Blade Element Momentum Model

The BEM model is comprised of two different yet related theories; the blade element theory and

the momentum theory [23]. Betz concluded that the BEM model works by breaking a lifting surface

10

into discrete elements and then calculating the effect that the rotating blade has on the flow at each

element. By monitoring the flow past each elemental lifting surface the forces on the rotor and power

generated by our turbine can be calculated. Glauert applied the angular momentum method, discussed

in Section 2.1.1.2, to concentric annuli that correspond with the existing Blade Element Model [27],

discussed in Section 2.1.1.1. Glauert also added an expression for the angular momentum balance, in

which changes in the angular momentum (from the free stream value of zero) are equated to the torque

exerted by the rotor on the fluid. This involves the introduction of a tangential or angular induction

velocity to manage the relationship between velocities at different locations [28].

2.1.1.1 Blade Element Model

The first portion of the BEM model is the blade element model which divides the blade into

multiple sections, elements or concentric annuli and calculates the forces on these elements as a

function of a lift coefficient, drag coefficient and angle of attack as demonstrated span-wise in Figure

4Error! Bookmark not defined. (a) and along the swept rotor plane area in Figure 4 (b) [29].

Figure 4: (a) Schematic of blade elements; c airfoil chord length, dr radial length of each element, r is the

distance from the radius, R is rotor radius andΩ is the angular velocity of the rotor [17]. (b) Schematic of the

blade elements with respect to the rotor plane area [22].

In order to do this, two assumptions must first be made [23]:

1. There is no aerodynamic interaction between the elements; in other words, the

spanwise flow is negligible which is true for large aspect ratio blades.

2. The forces on the blades are determined solely by the lift and drag characteristics of

the airfoil shape of the blades.

11

Figure 5: Local loads on a blade [17]

This theory eventually yields the two fundamental equations [23]:

( ( ) ( )) , (1)

where defines the differential normal/axial force (thrust) operating at a distance from the center

on a cord length of with number of blades at angle of relative water velocity operating on the

differential area and

( ( ) ( )) , (2)

where is defined as the differential tangential moment (torque) on the annular rotor sections as a

function of the flow angles at the blades [30].These two equations need to be numerically integrated

along the entire length of the blade in order to calculate the total drag (thrust) and torque on the rotor.

12

2.1.1.2 Momentum Model

Momentum theory takes an annular control volume approach that utilized the forces at the

blade to calculate the reduced incoming flow velocity based on the conservation of linear and angular

momentum [23]. This is achieved with the understanding that forces on the rotor blade and the flow

conditions at the rotor blade are related by the balance of momentum, since force is simply the rate of

change of momentum. By using the annular control volume the angular and axial induction factors can

be assumed to be a function of the radius [17]. In the momentum model some of the axial flow is

deflected away from the turbine which results in the flow past the rotor to have a velocity less than the

free stream velocity, the axial induction factor is the ratio of this velocity loss in the far-field.

By applying the conservation of linear momentum to the annular control volume of radius

and thickness the differential contribution of thrust can be expressed as suggested by [23],

( ) , (3)

where is the differential thrust, is the density of the fluid body, is the axial induction factor and

is the radial length. Similarly, the conservation of angular momentum equation allows the torque to

be calculated as suggested by [23],

( ) , (4)

where is the tangential induction factor and is the rotational velocity. Thus, thrust and

torque on each annular section of the rotor can be defined as a function of the axial and angular

induction factors or in other words, the flow conditions [23]. The tangential induction factor, also

known as the wake rotation, is the ratio of the rotational velocity that has been lost by passing through

the rotor and the free stream rotational velocity. Unfortunately the simple momentum theory alone

provides only an initial idea regarding the how well a propeller or turbine may perform, but does not

provide enough sufficient enough information to allow for detailed design [29].

2.1.2 Complete Blade Element Momentum Model

This section ties the previous two sub-sections, the Momentum model and the Blade Element

model, together to introduce the Blade Element Momentum model. These two models are then joined

13

together by incorporating the geometry of the into the annular control volume. The characteristics of

the blade such as local cord length, angle of attach and lift and drag coefficients are included in the

equations for force. The momentum It also introduces adjustments that have been made to the original

model to increase its accuracy. These include Prandtl’s tip loss factor and Glauert’s Correction for

High Values of .

Using the Blade Element model (Equations 1 and 2) and Momentum model (Equation 3 and

4) it is possible to calculate the steady thrust and power for different flow speeds, rotational speed, and

blade pitch angles – Equations 1-4 have four unknowns, , which is incorporated a , , as it relates

to , and , and therefore can be solved. Figure 6 shows the control volume approach

used with the BEM model [17].

Figure 6: Control Volume shaped as an annular element used with the BEM model [17]

It is important to note that the lateral boundary of these elements consists of streamlines and therefore

there is no flow across the elements. This leads to two assumptions that must be made for the annular

elements in the BEM model:

14

1. No radial dependency – what happens at one element cannot be felt by the others.

2. The force from the blades on the flow is constant in each annular element, which

corresponds to a rotor with an infinite number of blades [17].

In order to adapt the second assumption, a correction factor, known as Prandtl’s tip loss, was

introduced. Glauert’s correction factor was introduced to the BEM model to adjust for an assumption,

keep the momentum theory from breaking down in high axial and tangential induction factors, which

lead to flow reversals and causes the following derived relation from one-dimensional momentum

theory to become invalid.

( ) (5)

2.1.2.1 Prandtl tip loss

In order to improve the actuator disk model, which included infinite number of blades, so that

it is more applicable finite blade propeller or turbine Prandtl introduced the concept of a tip loss into

the work Betz’s was doing on the Blade Element model [17,31]. The focus was to show that when

approaching the blade tip of a real rotor the circulation exponentially tends to zero [32]. The most

commonly used version of Prandtl’s tip loss factor is an approximated form derived by Glauert [27].

This version of Prandtl’s correction factor is computed as:

( )

(6)

where

(7)

15

is the number of blades, is the total radius of the rotor, is the local radius and is the flow angle

[17]. The correction factor is then used with the equations for differential normal force, , and

torque, , as seen in the following equations:

( ) (8)

and

( ) . (9)

Then, by equalizing equations 7 and 3 along with 8 and 4 and applying the solidarity

equation,

( ) ( )

(10)

where denotes the number of blades, ( ) is the local chord length and is the radialposition of the

control volume, the equations for and can be derived as:

( )

(11)

and

( ) ( )

(12)

where is the normal load coefficient and is the tangential load coefficient. For more detail on this

process refer to [17].

This process accounts for the end effects of the rotor blade on the flow field and is utilized by

the momentum equations (Equations 8 and 9) [9].This correction factor directly impacts the induction

16

factors, which influence the velocity calculations and in turn has cascading effects all the way through

this theorem.

2.1.2.2 Glauert Correction for High Values of a

When the axial induction factor exceeds approximately 0.4 the velocity in wake becomes

negative and the simple momentum equation breaks down by invalidating relationships developed in

the one-dimensional momentum theory[17] .In order to avoid this case, Glauert developed and

empirical relationship between the axial induction factor and the thrust coefficient [23]. The steady

state solution allows for different empirical relations between the thrust coefficient and can be

generated to fit measurements. For the purposes of this thesis it can be expressed as [17]:

( )

( ( ) )

(13)

By using the differential normal force and torque calculated in Equations 8 and 9 along with

the altered value for and by plugging in the induction factor for turbulent or unsteady states ,

which is generally accepted as 0.2 [17] and solving for . This correction is necessary to eliminate

the numerical instability described earlier for high values of that occurs when the Glauert correction

is implemented in conjunction with the presence of tip losses [33].

2.1.3 Unsteady Blade Element Momentum Rotor Model

By adding some engineering models to the classic blade element momentum model presented

above, the unsteady blade element momentum rotor model was created to estimate the convergence of

the above theories towards their steady state values when rapid changes in environmental conditions

or operating parameters are experienced. The unsteady BEM mathematical rotor model used in this

thesis is based on the approach suggested by [9, 18], which is similar to the one presented in [17]. In

this model the effect that the rotor has on the incoming flow is calculated over set of mesh of evenly

spaced nodes with respect to azimuth angle and radial location and covers the swept area of the rotor.

This mesh is fixed with respect to the body fixed coordinate system. The forces acting on the rotor

blades are calculated using the momentum model with values interpolated from adjacent mesh grid

17

points. Conversely, the flow reductions at the mesh grid points used by the momentum model are

calculated as if a blade element were at each grid point. Therefore, both the actual rotor forces and the

rotor force calculations used for the momentum model are calculated each time step [9]. Using this

approach it is possible to implement the BEM model to flow fields that vary both in time and space.

This allows them model to incorporate the effects of a wave induced water velocities and current

shear, which vary over the swept area of the rotor blade. The main aspect to note when using this

approach is that it is an unsteady approach. The value that is being solved for represents the

induced velocities at varying angles of attack. It is important to note that these equations solve for

quasi-steady state values by using the prior time-step (after the first time step) calculated values of the

following equations are subsequently subtracted from the right-hand side of the equations when

updating new values for the induced velocity [17].

( )

| ( )|,

(14)

where is number of blades, is lift, is the angle between the plan of rotation and the relative

velocity, is Prandtl’s tip loss factor, is a vector of the undisturbed velocity fields parallel to the

rotor axis, is Glauert’s correction factor, is the unit vector in the direction of thrust. The value

or representan induced velocity normal to the rotor plane that is caused by the discontinuous

pressure drop across the rotor plane area.

( )

| ( )|,

(15)

where and represent the tangential induced velocity generated due to the same cause.

This can be done since the induced velocity changes relatively slowly in time due to the

dynamic wake model. A dynamic inflow model is applied to Equations 12 and 13 to account for the

time delay before they are in equilibrium with the aerodynamic loads. The implemented model is a

filter for the induced velocities and is comprised of two first order differential equations [17].

18

(16)

where is the quasi-steady value found in Equations 12 and 13 and is an intermediate value.

(17)

The value of is then the final filtered value to be used as the induced velocities and for more detail

in the two times constants and see [17].

2.2 OrcaFlex

While the Ocean Current Turbine simulation presented by [9] accounts for waves and current

shear, it is somewhat limited by its numeric cable model, which significantly slows the simulation

down when more than 5 cable elements are utilized. To allow device developers to more easily

numerically model OCT systems that utilize complex mooring system, the rotor model utilized by [9]

is modified and then implemented into the OrcaFlex numeric modeling program as part of this thesis.

OrcaFlex is one of the leading software packages for the dynamic analysis of offshore marine systems,

specializing in mooring systems, buoys and vessels [34]. The OrcaFlex environment allows users to

easily set and adjust their simulated operating environment and model properties when evaluating

their systems. Unfortunately, with only a few basic rigid body buoy models available to the user, only

very basic models of OCTs created by matching the mass and drag of the buoy to the desired turbine.

OrcaFlex does allow for wings to be applied to the buoy that could provide basic rotor estimates, such

as those presented in [19, 20]. However, it does not model the impact that the rotor blades have on the

incoming flow, often resulting in a significant over estimation of the forces on the rotor and the power

produced by the turbine.

The following subsections provide background information on the OrcaFlex software that is

directly relevant to implementing BEM based rotor models into the OrcaFlex environment. Sub-

section 2.2.1 provides an overview of OrcaFlex’s coordinate systems, the objects used to build models

19

in OrcaFlex (i.e. 3-DOF buoy, vessel, 6-DOF buoy and cables), hereafter referred to as Elements,

(sub-section 2.2.2), external functions in sub-section 2.2.3, followed by a brief introduction into the

utilized equations of motion in sub-section 2.2.4 and the equations that are used to calculate the

hydrodynamic forces (sub-section 2.2.5). Specific attributes of external functions are discussed in

detail in two sub-sections: the first focuses on their Application Programming Interface (API) (sub-

section 2.2.3.1) and the second addresses variables that can be passed between the OrcaFlex

environment and the external function (sub-section 2.2.3.2). The of equations motion that used by

OrcaFlex including the external function impacts are then addressed (sub-section 2.2.4) followed by

an explanation of the equations that are imported into OrcaFlex from the external function (sub-

section 2.2.5).

2.2.1 OrcaFlex Coordinate Systems

OrcaFlex uses a combination of coordinate systems that allows it to perform its operations

effectively. The first is the earth fixed frame coordinate (EFF) system G , where G is the global

origin and , and are the global axes directions. OrcaFlex also utilizes local coordinate, also

known as body fixed frames (BFF), for each object include (i.e. each buoy or vessel) [35].

20

Figure 7: Image of OrcaFlex’s Coordinate systems [35]

These coordinate systems, seen in Figure 7, follow the right hand rule with Z positive

upwards (opposing gravity) for the EFF. The directions of the and coordinates can be chosen by

the user with the constraint that they lie in the horizontal plane and complete the right hand rule (i.e. if

is towards the north than will be towards the west).The BFFs for each type of object are described

by the OrcaFlex Manual in the section about that object, but typically the origin is at a selected fixed

point on the object and the axes are in special fixed directions, such as the surge (forwards), , sway

(towards Port), , and heave (upwards), , directions for a vessel and a 6D buoy [35].

2.2.2 Elements

OrcaFlex provides a wide range of elements the user can manipulate. Listed in order from the

GUI window these elements are: vessels, lines/cables, 6-DOF buoy, 3-DOF buoy, winch, new link and

new shape [34]. Each of these elements has a property pane that can be viewed when double clicked

through the model browser or on the object itself in the GUI window (Figure 8).These elements can be

tailored to fit a multitude of uses to match the users demand. For example the 6-DOF buoy has three

different options for their design type which includes: Spar buoy, lumped parameter buoy and towed

fish.

21

In order to simulate SNMREC’s experimental OCT, a main foci of this thesis, the model

could only consist of elements that were provided by OrcaFlex’s API. Even though the external

function is designed to use the programming interface, the elements that make up the model were

designed in the GUI in order to more clearly visualize the modeling process as well as the location of

the nodes of the mooring cables. The first step of the model building process is the selection of an

OrcaFlex element to act as the turbine. A 6-DOF buoy is selected because it is the only element

OrcaFlex has that could provide all of the information necessary to model the turbine. Using the 6-

DOF buoy provides the necessary state variables each time step to the external function, which allows

for modeling the influence of environmental conditions on the OCT. In order to model the complete

OCT system a mooring system also needed to be attached, much like the cable model in [18].

OrcaFlex allows mooring line elements to be easily attached to the buoy using the GUI and the forces

applied to the cable element from the 6-DOF buoy can be updated by the external function. This

Figure 8: Screen Shot of OrcaFlex’s Graphic User Interface.

22

allows the thrust created by the turbine and the orientation of the turbine to affect the stress and

orientation of the mooring line and demonstrate the coupled effects.

2.2.3 External Functions

OrcaFlex can be used as a normal Windows Graphic User Interface (GUI) program (Figure 8)

or driven via a programming interface, or both [34]. The programming interface allows the user to

access and use a limited range of OrcaFlex's facilities to better suit individual user needs. The

implementation of external functions, which is done through a programming interface, provides the

means for users to create custom numeric models that suit their needs when the standard calculations

performed by OrcaFlex are inadequate. In order to incorporate the BEM model it is necessary to

receive variable updates each time step to accurately account for the momentum effects on the flow

and the blades, while incorporating the impact of the BEM calculated rotor forces into the response of

the mooring system. Knowing which values are utilized for these calculations is important because it

limits the types of elements that can be used in the GUI model, as well as the programming language

that the external function utilizes. In order to collect the necessary variables each time step the

program must either be written in Delphi or an object oriented C derivative, such as C or C++ [34].It

is important to note that the external function must be written in a way that allows for the creation of a

Dynamic-Link Library (.dll) file when built, so that external function could interact with OrcaFlex’s

Application Programming Interface (API).

2.2.3.1 API

In order to interface between different applications on any operating system it is crucial to be

familiar with the API of both applications. The API is essentially a list of the variables, routines, data

structures and object classes specific to each application. In this particular application, this means that

in order to access the variables that OrcaFlex updates, the external function must call from the data

structure that OrcaFlex stores them in using the same format that OrcaFlex uses to call them. For

example, OrcaFlex’s API stores the majority of its variables inside data structures for a wide array of

variable types and uses Long Pointers to Constant TCharSTRings (LPCTSTR) for most of its function

23

calls. This is due to the fact that Windows has moved away from the 2-bit character encoding and into

Unicode Transformation Format 16 bit (UTF-16) character encoding.

2.2.3.2 Variables

As previously mentioned the majority of the variables provided by OrcaFlex’s API are stored

within data structures, most of which are grouped by the element types described in subsection 2.2.2.

This allows variables of similar nature to be more easily accessed. In the case of the 6-DOF buoy that

is used in the simulation of SNMREC’s experimental OCT, the API provides access to a structure

denoted TBuoyInstantaneousCalculationData which holds the following variables [34]:

a position vector (buoy origin relative to the global origin),

orientation matrix (orientation of the buoy, relative to global axes),

vector velocity (velocity of the buoy origin relative to global),

vector angular velocity (angular velocity of the buoy relative to global),

vector for the wetted centroid location (local axes coordinates of the position of the

centroid of the wetted portion of the buoy),

vector for the undisturbed lumped fluid velocity at the wetted centroid position

(velocity of the fluid, current and wave combined, relative to global axes).

This fixed number of variables limits the calculations that can be performed in the external

functions and the numeric models that can be utilized. The undisturbed fluid velocity that is provided

by OrcaFlex combines the current velocity along with the wave velocity at the wetted centroid and

does not provide a profile for the swept water plane area.

2.2.4 Equations of motion

To properly account for the 7-DOFs, pitch, roll yaw, x, y, z and the rotor rotation, of an OCT

with a single rotor, that is modeled as having two rigid bodies allowed to rotate relative to each other

about a shaft at a pre-determined rotational velocity, it is important to document how OrcaFlex

24

handles the equations of motion for the 6-DOF buoy, which are being utilized when modeling the

OCT. OrcaFlex offers two dynamic integration schemes, Implicit and explicit, the latter is used for

this thesis so that a constant time step can be utilized. The explicit scheme utilizes the forward Euler’s

method for a fixed time step. To calculate the acceleration of a standard 6-DOF OrcaFlex element the

following equation is solved:

( ) ( ) ( ) ( ) (18)

where ( ) is the system inertial load, ( ) is the system damping load, ( ) is the system

stiffness load, ( ) is the external load, is the position and attitude vector, is the relative

linear and angular water velocity vector, is the linear and angular acceleration vector and is the

simulation time. This relationship is utilized to calculate the linear and angular accelerations of the 6-

DOF element, which are then numerically integrated to update the states of the system (Equation 17).

OrcaFlex considers the forces and moments from gravity, buoyancy, hydrodynamic and aerodynamic

drag, hydrodynamic added mass effects (which are calculated using the extended form of Morison’s

Equation with user defined coefficients), tension, shear, bending, torque, seabed reaction (including

friction) and contact forces with other objects [35].

To solve for the accelerations the linear and angular acceleration vector is factored out of the

inertial load matrix using Newton’s Second law:

( ) ( ) ( ) ( ) (19)

where ( ) is a 6x6 mass matrix comprised of the mass and moments of inertia, in the body fixed

coordinate system in the following form [36]:

25

( )

[

]

.

(20)

This is the local equation of motion used for each 6-DOF buoy and is not the same as Equation 16

since the acceleration is now an independent variable. In order to apply the forces and moments

calculated by the external function an addition term can be included in Equation 17 yielding:

( ) ( ) ( ) ( ) ( ) (21)

where ( ) is the force and moment vector calculated by the external function. In order to solve

this equation, all that is required is the inversion of the 6-DOF inertial matrix.

( )[ ( ) ( ) ( ) ( )] (22)

At the beginning of each time step this equation is solved for the acceleration of the 6-DOF buoy,

which is then integrated using Euler integration. At the end of each time step the position and

orientation are updated and the process repeats [35]. In this equation the moments are calculated about

the origin of the body fixed frame and the linear velocities are those of the origin of the body fixed

frame.

2.2.5 External Buoy Forces

To numerically model an OCT in OrcaFlex the external forces, ( ), on the turbine

utilized in the development of the OCT (chapter 3) are discussed here. The following equations are

used by the standard OrcaFlex external force equations available in the 6-DOF buoy element. These

include those induced by weight, buoyancy, hydrodynamic and aerodynamic drag, hydrodynamic

added mass effects (hydrodynamic drag and added mass are calculated using the extended form of

Morison’s Equation with user defined coefficients) [35]. In this thesis the force and moment vectors

are denoted as those from: weight, , buoyancy, , and the hydrodynamic interactions (both drag

26

and added mass) calculated using Morison’s Equation, ; which can be summed to calculate the net

external force and moment vector via:

( ) . (23)

Before delving into the force equations it is important to clarify how and where OrcaFlex

applies the calculated forces, when calculating the moments about the origin of the body fixed frame.

The 6-DOF buoy used to model the turbine is treated as a rigid body, with 3 translational and 3

rotational degrees of freedom. The hydrodynamic forces on 6-DOF buoys are calculated and applied

to the center of volume, where the center of volume is defined as:

( )

for a fully submerged buoy, (24)

where is the center of wetted volume, is the proportion that is wet and is the height of the

buoy. The moment on the 6-DOF buoy is then calculated using this force vector and the distance

vector between the center of volume and the origin of the body fixed coordinate system. It is

important to note that various components, including other 6-DOF can be rigidly linked to each other

to model the external forces and moments on complex shapes.

Weight

Calculation for weight forces and moments on 6-DOF buoys requires the user to define the

mass of the buoy and the distance vector from the origin to the center of mass. This is necessary for all

of the components that are rigidly linked together to create the modeled rigid system. This value needs

to be calculated before running the simulation as OrcaFlex provides the user with three input boxes in

the GUI window for each element that is defined, one for each of the directions in the body fixed

coordinate system. The first equation in this section addresses the weight of the 6-DOF buoy. The

weight force equation is represented by:

27

[ ] [

],

(25)

where is OrcaFlex’s transformation matrix, is the total buoy mass that is defined by the user

and is the gravitational constant. The moment caused by the weight force is therefore calculated

from:

[

] ,

(26)

where denotes the cross product.

Buoyancy

For the lumped buoy the buoyancy force is given by:

[ ] [

], (27)

where is the buoyancy, is the sea density and is the wetted volume. The buoyant force is then

applied vertically at the center of the wetted volume, . The moment created by the buoyant force is

therefore calculated from:

[

] .

(28)

where and represent the , and locations of the center of wetted volume.

Hydrodynamic Loads and Moments

In OrcaFlex the hydrodynamic loads are calculated and applied at each buoy’s and element’s center of

volume. These forces are calculated using the extended form of the Morrison Equation:

28

( )

⁄ | |, (29)

where is the fluid force vector, is the mass of the fluid displaced by the body, is the fluid

acceleration relative to the earth, is the added mass coefficient for the body, is the fluid

acceleration relative to the body, is the drag coefficient for the body, is the drag area and is the

relative water velocity vector. This equation is comprised inertial (portion on parentheses) and drag

components, with the inertial component including the Froude-Krylov force and added mass term.

These forces, and their distance from the origin, can be used to calculate the hydrodynamic moments

about the origin by:

[

] .

(30)

OrcaFlex elements can be set a fixed distances and orientations with respect to the origin of the body

fixed frame. Each individual body’s forces, and the moments about the common body fixed frame are

summed to calculate the net hydrodynamic forces and moments on the buoy.

.

29

3 OCT MODEL DEVELOPMENT

In this chapter the mathematical models utilized to predict OCT performance are presented.

This chapter focuses on the numerical calculations developed/implemented as part of this thesis, and

not those available as part of the standard OrcaFlex package (these are summarized in Section 2.2). It

also does not focus on the environmental conditions utilized to evaluate the numerically simulated

performance, as these are summarized with the individual analyses presented in Chapter 4. The first

section (Section 3.1) discusses the coordinate systems and kinematics used in modeling the motions of

an OCT, without referencing the force that causes the motion. The second section (Section 3.2)

provides a description of the equations of motion that are utilized in the overall behavior of the

system. The next section in this chapter (Section 3.3) describes how the hydrodynamic forces on the

on the OCT are calculated. The final section (Section 3.4) addresses specific properties of the

components of the OCT that is used for this thesis.

3.1 Coordinate Systems and Kinematics

The developed simulation will run within the OrcaFlex environment and therefore their EFF

and BFF coordinate systems will be utilized in the calculations (see section 2.2.1). Additionally, the

three coordinate systems utilized to calculate the forces on the rotor, presented later in this section,

will utilize the conventions described in [9]. These five coordinate systems are defined as: the earth

fixed coordinate system, ; the body fixed coordinate system, ; the momentum mesh coordinate

systems, , where ( ) indicates the referenced blade element radial location on mesh azimuth

angle grid point ( ) ; the shaft coordinate system, ; and the rotor blade coordinate systems,

,

where ( ) indicates the referenced blade element on the rotor blade ( ) . The ( ) symbol denotes all

potential variables that use the above listed superscripts. For the numerical simulations presented in

this thesis the earth fixed coordinate system is located at mean sea level; with the -axis oriented

30

north, the -axis oriented west and the positive -axis is upward. The body coordinate system is

positioned at the center of the rotor shaft directly behind the pressure vessel; with the -axis aligned

with the rotor shaft and running from tail to nose, the -axis aligned with the port direction, and the -

axis towards the top if the turbine.

Figure 9: Lumped buoy coordinate system. Image from [35]

In order to utilize a polar coordinate system that covers the entire swept rotor plane area and

is centered at the center of the rotor hub, each is attached to a discrete mesh point that is fixed with

respect to . Each coordinate system is comprised of an axial directional component ( ) which is

parallel to the -axis, a tangential directional component ( ) that follows the rotor rotation direction

and a radial component ( ) that is oriented outward in the radial direction from the center of the

rotor. The mesh is divided into angular sections spaced evenly with respect to the azimuth angle,

with the angle component of the matrix denoted by ( ) .

The origin of the shaft coordinate system is fixed to the shaft at the center of the hub and

rotates with the rotor. For this coordinate system, the -axis is co-axial with the body-fixed -axis,

the -axis is set to be perpendicular to the rotor shaft and parallel to rotor blade 1 (an arbitrarily

31

chosen rotor blade), but extending from the shaft in the opposite direction and the -axis is aligned to

complete the right hand rule. In the final coordinate system, each

is fixed to the quarter cord line of

each of the discrete rotor blade sections and is comprised of the axial directional component ( )

which is aligned parallel to the -axis; a tangential directional component ( )

oriented in the

direction rotor rotation and a radial direction component ( )

that points outward in the radial

direction from the rotor’s center.

3.1.1 Transformation Matrices

In order to shift between coordinate systems it is necessary to implement transformation

matrices. Since OrcaFlex handles the coordinate systems and it is not necessary to do the

transformation matrix calculation for this externally. These differ from the transformation matrices

presented in [9] because OrcaFlex uses the bow, port and up as the positive directions in their body

fixed coordinate system while the model presented in [9] uses bow, starboard and down as the positive

directions in their body fixed coordinate system:

[

] [

] [

]

(31)

In order to transition variable from to , the body fixed coordinate system to the mesh

coordinate system, a constant transformation matrix is utilized. This matrix is defined as:

[

]

(32)

where ( ) is the azimuth angle of grid point( ) .

To transition variables from to the relative angle between the rotor and the turbine, ,

is utilized and this transformation matrix is defined as:

32

[

] (33)

Finally, the constant transformation matrix that converts to

is defined as:

[

]

(34)

where is defined as the angle between the reference rotor blade (blade 1) and the rotor blade of

interest.

3.2 Equations of Motion

The general 6-DOF rigid body equations of motion, as presented in [36], are represented as:

[ ( ) ( ) ( )] (35)

[ ( ) ( ) ( )] (36)

[ ( ) ( ) ( )] (37)

( ) ( ) ( ) ( )

[ ( ) ( )]

(38)

( ) ( ) ( ) ( )

[ ( ) ( )]

(39)

( ) ( ) ( ) ( )

[ ( ) ( )]

(40)

where denotes mass, the linear velocity in the -direction, the linear velocity in the -direction,

the linear velocity in the -direction, the rotational velocity in the -direction, the rotational

velocity in the -direction, the rotational velocity in the -direction, the center of gravity in the -

33

direction, the center of gravity in the -direction, the center of gravity in the -direction, is

the moment of inertia for the -direction, is the moment of inertia for the -direction, is the

moment of inertia for the -direction, is the product of inertia for the plane, is the

product of inertia for the plane, is the product of inertia for the plane. The rotation of

the rotor is then incorporated into the rotational velocity of the system as in [37]. Those equations

were then applied to each of the rigid bodies, the rotor and the main body, resulting in the following

equations:

[ ( )

( ) ( )] [

(( ) )

(( ) )]

(41)

[ (

) ( )

( )]

[ ( ) (

) (( ) )]

(42)

[ (

) ( )

( )] [ ( )

(

) ( ( ) )

(43)

( ) ( ) ( ) ( )

[ ( )

( )] ( ( ) )

(( ) ) [ ( ( ))

( ( ) )]

(44)

( ) ( ) ( ) ( )

[ ( ) ( )] ( ) ( )

( ) ( ( ) ) ( (

))

(45)

34

( ) ( ) ( ) ( )

[ ( )

( )] ( )( )

( ( )) ( )

( ( )

)

(46)

where is the mass of the body of the turbine, is the mass of the rotor, is the center of gravity

of the turbine body in the -direction, is the center of gravity of the turbine body in the -

direction, is the center of gravity of the turbine body in the -direction,

is the center of gravity

of the rotor in the -direction, is the center of gravity of the rotor in the -direction,

is the center

of gravity of the rotor in the -direction, is the rotational velocity of the turbine body and the

rotational velocity of the rotor, which are only valid at a constant RPM, that incorporates the

gyroscopic forces. The equations are finally in their full form for our system, now they will be reduced

to their simplest form using properties of the buoys, coordinate systems and assumptions. For these

calculations, the center of gravity was placed at the center of the rotor. With that assumption, , ,

, ,

and are all set equal to zero and the subscript and superscript can be dropped and

viewed as totals like: ,

and . Those assumptions lead to the development

of the following equations:

[ ( ) ( ) ( )] (47)

[ ( ) ( ) ( )] (48)

[ ( ) ( ) ( )] (49)

( ) ( ) ( ) ( )

[ ( ) ( )]

(50)

35

( ) ( ) ( ) ( )

[ ( ) ( )] (

)

(51)

( ) ( ) ( ) ( )

[ ( ) ( )] (

)

(52)

There is now an additional term on the right hand side of the equation that incorporates gyroscopic

forces created by the rotating turbine. OrcaFlex does not have a direct manner for handling this term

so the will be added to the ( ) , the forces and moments vector calculated by the external function

as a result of the rotor forces on the system, term calculated within the developed external function so

that the equations of motion discussed in Section 2.2.4 can be utilized without neglecting these

additional forces and moments.

3.3 Hydrodynamic Modeling

This section presents the approach utilized to calculate the hydrodynamic forces and moments

on the OCT system. Section 3.3.1 introduces the calculations utilized to calculate the forces on the

rotor, which are implemented into OrcaFlex using an external function. The Dynamic Wake model

(Section 3.3.1.1) is used to model the dynamic impact of the rotor on the incoming fluid. In the next

section (Section 3.3.2), the approach utilized to calculate the hydrodynamic forces and moments on

the remaining components using standard OrcaFlex elements is documented.

3.3.1 Rotor Modeling

In this section the mathematical rotor modeling techniques utilized in this thesis are discussed

in detail. This model utilizes an unsteady form of the BEM model to calculate the forces on the rotor

blades. The inputs into this rotor model are the OrcaFlex provided states (Section 2.2.3) and the

momentum states calculated from previous time steps. From these states the rotor model updates the

momentum states, which are utilized during the subsequent time step, and calculates and net

hydrodynamic forces and moments on the rotor, which are exported to OrcaFlex.

36

This model calculates the momentum loss in the flow field caused by the rotor forces in the

coordinate system. The impeded flow at each rotor blade element is then interpolated from the

adjacent radial points on this momentum mesh grid when calculating the relative water velocity

encountered by each rotor blade. Conversely, the flow reduction caused by the momentum loss is

calculated as if a blade element were at each momentum grid point. Therefore, both the actual rotor

forces and the rotor force calculations used in the momentum model are calculated each time step.

For each of the discrete radial location ( ) of both the rotor blade ( ) and azimuth angle of

the mesh grid ( ) , the angle of attack is calculated as a function of the axial

and

tangential

components of relative water velocity ,

[

]

.

These relative water velocities are calculated from:

(53)

( ). (54)

In this equation

is the effect of the device motions on the relative water velocity (from

Equations 22 and 23),

represents the combined velocity of the undisturbed free stream velocity

from both the waves and current (from Equations 25 and 26), and ( ) is the wake induced

velocity calculated from the previous time step. ( ) is initialized as a zeros matrix and is

then calculated using Equation 50 (Section 3.3.1.1) in subsequent time steps.

The

in Equation 21 is calculated from

[[

] [

] [

]] and

(55)

37

[[

] [

] [

]]

(56)

where and are the angular velocities about the and -axes imported from OrcaFlex and is sum

of the relative rotational velocity of the rotor about the -axis and the rotational velocity of the turbine

about the -axis imported from OrcaFlex:

(57)

The

term in Equation 21 is calculated from

[

]

[

] and

(58)

[

] [

]

(59)

where the undisturbed velocities imported from OrcaFlex in ( ) are functions of the

current profile, wave field, the turbine shaft location and time. The three scalar values, , and

, are imported from OrcaFlex each time step and are not dependent radial location ( ) , but are

made to be dimensionally consistent with the other matrices.

To calculate the hydrodynamic forces on the rotor, the angle of attack of each section of the

blade, is calculated by

(60)

where the relative flow angle in

is calculated from

38

(

) (61)

and is the blade section pitch angle, which is only a function of radial location.

Using the angles of attack calculated in Equation 26 with both the lift and drag coefficient

matrices, ( ) and

( ) respectively, the axial and tangential force coefficients are calculate

by:

( )

( ) and (62)

( )

( ). (63)

Using these coefficients axial and tangential loads on each of the blade sections are estimated by:

((

) (

) ) and

(64)

((

) (

) ),

(65)

where is the density of sea water and is the cord length at the center of center of section ( ) .

These forces are converted to by

[

]

[

]

(66)

and summed to calculate the total forces on the rotor

[

] ∑∑[

]

(67)

39

where ( ) is used in reference to the entirety of the rotor blade. Similarly, the hydrodynamic moment

from the rotor, with constant RPM, about the origin is calculated from

[

] ∑ ∑ [

] [

]

.

(68)

The forces calculated in Equation 33 and the moments calculated in 34, are summed with the

gyroscopic terms on the right had side of equation 41-46 to calculate the forces and moments exported

from the external function,

( )

[

]

[

(

)

(

) ]

(69)

to the OrcaFlex environment where they are summed with the forces and moments calculated directly

by OrcaFlex (Equation 20).

3.3.1.1 Dynamic Wake Model

This section describes the unsteady portion of the BEM model utilized to update the

calculated impact of the rotor on the incoming flow field. This requires calculating the quasi-static

impact of the rotor forces on the incoming flow field and then numerically converging the calculated

flow field to this constantly changing quasi-static value using a first order differential equation. This

section addresses the means by which a time delay is implemented and the effects the spinning rotor

blade has on the wake flowing past.

As mentioned earlier there is tip loss effect that has not been accounted for yet since the flow

past the coordinate system covers the entire swept rotor plane area, not accounting for the

individual blades. In order to make adjustments for the end effects of the rotor on the flow field

40

Prandtl’s tip loss factor, like that presented in Section 2.1.2.1, is applied. The tip loss factor used for

this simulation is calculated by

(

( )

( )),

(70)

where is the number of blades and is the total radius of the rotor.

For the time elapsed wake field, the updated axial induction factor from the previous time

step is defined as

( )

‖ ‖

(71)

Where ‖ ‖ denote the or Euclidean norm. Using this induction factor, the Glauert empirical

correction factor is calculated using

(

)

(72)

where .

The written external function allows device specific 3-D lift coefficients matrices, written as a

function of angle of attack and radial location, to be called at each time step. The lift coefficient

matrix, ( ) is calculated at each mesh node as a function of angle of attack, found in (27), in

order to obtain the lift per unit length,

(( )

(

) ), (73)

where is the radial distance from the rotor shaft elements and is the chord length at each radial

location.

41

The quasi-static wake field is now calculated in terms of its axial and tangential components

for time step :

( ) ( )

√(

( )) (

) (

)

(74)

( ) ( )

√(

( )) (

) (

)

(75)

where ( ) denotes that the utilized wake field is not corrected for the wake skew angle in Equation

38 and 39 and is the lift per unit length calculated in Equation 38.

Following the method provided by S. Oye, a filter is applied that consists of two first order

differential equations, Equations 26 and 27. These differential equations are solved analytically by

using intermediate wake variable vectors and as follows:

( )

( )

( )

(76)

( ) (

( ) ) (

)

(77)

( )

( ) ( ( )

( )) (78)

These equations are reliant on the constants and which are calculated by:

(79)

( )

(80)

42

( (

)

) (81)

This method is used because it calculates the impeded flow velocity for discrete locations over the

swept rotor plane area with a discrete mesh fixed with respect to the . To estimate the wake at the

blade elements, ( ), the wake is linearly interpolated between the nearest two azimuth angles in

the mesh grid ( ) for the same radial location:

( ) (

( ) ) (82)

A yaw model is included in this simulation that accounts for the relative yaw angle of the

rotor when calculating the impact of the rotor on the surrounding flow field. The method proposed by

Glauert is used to calculate this wake field corrected by

( ) ( )(

(

) ( ))

(83)

where is the wake skew angle defined as the angle between the current velocity in the wake and the

rotational axis of the rotor and is the angle where the blade is deepest into the wake. This wake

field is fed back into the relative flow speed calculations, Equation 21, in the following time step. The

skew angle in this equation is defined by

(‖∑(

) ∑(

)

∑(

)

)

(84)

where ( ) denotes the four quadrant inverse tangent function and ( ) denotes that the skew

angle is assumed to be constant with the radius and is calculated at

⁄ = 7 as suggested by [17].

43

3.3.2 Streamlined Body Forces

The methodologies utilized to calculate the hydrodynamic forces on the streamline bodies will

follow those suggested by [20]. Below is a brief summary of this approach, including the fundamental

equations utilized to calculate the forces and associated moments. The streamline body forces that are

mathematically modeled using this approach include: two separate sections of the nacelle (the main

body of the turbine) and two Buoyancy Compensation Modules (BCM). The nacelle sections and

BCMs are modeled using standard OrcaFlex pipe elements. The drag forces and resulting moments

about the origin of the body fixed frame are calculated using the relative axial and tangential water

velocities calculated at the center of volume of each of these four bodies:

[

]

(85)

and

[ ] [

] (86)

where, is the mass of the fluid displaced by the body, is the fluid acceleration relative to the

earth, is the normal drag coefficient, is the coefficient of axial drag, is the frontal projected

area of the body, is the average diameter of the streamline body, is the length of the

streamline body, is the relative section speed through the water [8]. The forces and moment vectors

from these streamline bodies (two from the nacelle and one from each of the two BCMs) are summed

by OrcaFlex to calculate the total hydrodynamic force and moment vector:

[

],

(87)

which are summed with the other forces and moments according to (Equation 21) to calculate the total

force on the turbine.

44

3.4 Properties of SNMREC’s OCT

This section discusses the characteristics of the nearly neutrally buoyant SNMREC OCT

design that is utilized for creating the example results presented in this thesis. In order to compare the

OrcaFlex predictions with previously published results, the simulated system is based on the same

prototype design modeled by [20, 9]. The inertial, buoyancy, hydrodynamic, and geometric properties

of this turbine are presented in this section in the format necessary so that they can be directly

implemented into the external function and OrcaFlex 6-DOF buoy. These include the mass and inertial

system properties (Section 3.4.1) and hydrodynamic properties (Section 3.4.2).

3.4.1 Mass, Displace Volume and Inertial Properties

The mass, displaced volume and inertial properties of the modeled system are presented in

this section. These properties are derived from the nearly neutrally buoyant version of SNMREC’s

OCT presented in [9, 20]. The mass and displaced volumes of each object in this turbine design, as

well as the relative locations, are utilized for these calculations. The values in Table 3 are calculated

from this turbine design and presented in the form needed for implementation into OrcaFlex. Since

OrcaFlex allows users to implement the inertial properties directly into the program, these can be

directly calculated. However, OrcaFlex bases its buoyancy calculations on the volume and center of

volume of its various components and therefore needed to be calculated separately for the rotor, two

nacelle components, and two BCMs. Also, to allow for the placement of the BCMs to be easily

changed to optimize performance, the mass, volume and inertial properties of these are accounted for

separately from the rest of the system. The values in this table are given about the origin of the body

fixed frame, which is located at center of mass of the rotor.

45

Table 3: System mass, centers of Gravity, and mass moments with respect to the center of mass of the rotor and

the center of buoyancy that are plugged in to OrcaFlex(Products of inertia for x-y and y-z plane are zero due to

symmetry of OCT).

The values that are used by OrcaFlex (Table 3) correspond to an OCT system that that has the

mass and buoyancy characteristics that are presented in Table 4.

Table 4: Inertial characteristics of each component of the OCT system.

The turbine is almost neutrally buoyant with a net buoyant force of 302.82 N. The center of buoyancy

of this system is almost directly above the center of gravity, with the center of buoyancy 0.91 cm in

front of the center of gravity and 56.45 cm above the center of gravity.

3.4.2 Hydrodynamic Properties

The majority of the hydrodynamic system properties presented in this portion of thesis follow

those utilized by [9, 20]. In the OrcaFlex simulation used to model SNMREC’s OCT there are 3

Mass 1543.10 kg Length 2.34 m Length 0.508 m

Center of Gravity (x) 1.44 m Diameter 0.48 m Diameter 0.45 m

Center of Gravity (y) 0 m Density 162.89 kg/m3

Density 0.00 kg/m3

Center of Gravity (z) 0.15 m

Center of Buoyancy (x) 1.17 m Length 2.80 m X-Coordinate 3.03 m

Center of Buoyancy (y) 0.00 m Diameter 0.55 m Y-Coordinate 0.00 m

Center of Buoyancy (z) 0.04 m Density 0.00 kg/m3

Z-Coordinate -0.05 m

Mass Moment of Inertia (x) 337.51 kg-m2

Mass Moment of Inertia (y) 4435.60 kg-m2

Mass Moment of Inertia (z) 4187.00 kg-m2

Product of Inertia (x-z Plane) 408.47 kg-m2

Entire System (Not including BCMs)

Nacelle (Forward)

Nacelle (Aft)

Cable Attachment Point

BCMs

46

components that are hydrodynamically modeled. These modeled components are the rotor (section

3.5.2.1), the turbine nacelle (section 3.5.2.2) and the buoyancy compensation modules (section

3.5.2.3).

3.4.2.1 Rotor Hydrodynamic Properties

The hydrofoil shape of the numerically modeled rotor blades range in shape from

approximately NACA 4412 at the hub to NACA 4421 at the tip and transition through NACA 4415

and NACA 4418 between the hub and tip [9]. This OCT is designed to operate in a flow that is

approximately 2.0 ⁄ , with the rotor designed to operate near a Tip Speed Ratio (TSR) of roughly

4.0. XFoil, a tool designed by MIT, was used to calculate the 2D lift and drag coefficients for these

four airfoil shapes that are approximately evenly spaced from hub to tip by. The hydrofoil properties

are then entered into Airfoil Prep, a tool designed by NREL, to calculate the 3D lift and drag

coefficients for these four airfoil cross-sections. The method used to transition these four airfoil data

sets into lift and drag coefficients into a user specified number of evenly spaced airfoil cross-section

from hub to tip used a custom developed polynomial curve fitting routine, along with the sixty-four

predefined angles of attack, to create a set of second order polynomial equation (one for each pre-

defined angle of attack). These polynomials are then utilized, along with a linear interpolation routine,

to calculate the corresponding lift and drag coefficient for any angle of attack and radial location. The

lift and drag coefficients for 100 evenly spaced airfoil sections are plotted as a function of angle of

attack in Figure 10, with a zoomed in version of this figure presented in Figure 11.

47

Figure 10: 3D Lift and Drag Coefficient as angle a function of attack for all possible angles of attack

Figure 11: 3D Lift and Drag Coefficients as a function of angle of attack over the expected operating conditions

3.4.2.2 Turbine Nacelle

The hydrodynamic model used to estimate the forces on the turbine nacelle (the main body of

the turbine) is comprised two OrcaFlex pipe element that is separated into two distinct sections with

slightly different characteristics. The aft section of the nacelle, which extends in the negative

48

-direction behind the rotor, has a diameter of 0.508 m and a length of 0.508 m, while the forward

section of the nacelle has a diameter of 0.509 m and a length of 2.805 m [20].

OrcaFlex defines common values to act as default values when using the OrcaFlex bodies,

6-DOF buoys, lines/pipes, etc. For the majority of line or pipe elements, the forces at the end of the

pipes is typically neglected due to most of the implementations involve pipes of extensive length

which causes end drag to play minimal role when compared to skin drag. However, for our application

the length of the pipe is relatively short and therefore the end drag must be included. For this thesis the

default normal drag coefficients suggested by OrcaFlex for both of these sections are =1.2 is used.

The axial coefficients are a combination of end and skin drag for the back and front sections.

Assuming a skin drag coefficient of 0.008 and an end drag coefficient of 0.6 the coefficients

implemented into OrcaFlex and are calculated as follows:

(

) (

)

(88)

(

) (

)

(89)

where the skin drag coefficient is represented by , the end drag coefficient is , the resulting back

axial coefficient utilized by OrcaFlex is , and the resulting front end axial coefficient utilized by

OrcaFlex is . The added mass for the main body also uses the default value for a cylinder for both

axial ( ) and normal added mass (

) [20].

3.4.2.3 Buoyancy Compensation Modules

The buoyancy compensation modules on the numerically modeled version of the SNMREC

OCT are modeled as two separate ellipsoid bodies, each one evenly spaced with a center of volume

located 0.432 m, left and right of the rotor hub, 1.7 m above the rotor hub and 0.57 m forward of the

rotor plane area. For modeling simplicity, each of the BCM was given a constant diameter of 0.482 m

49

and length of 2.337 m in order to match the buoyancy of the actual specifications, which has the same

length of 2.337 m but has a varied diameter with a maximum of 0.53 m.

Since the BCM is modeled as a pipe in OrcaFlex as well, OrcaFlex’s default of for

normal drag coefficient was selected. The skin drag and end drag, are then used to calculate a

combined axial drag coefficient by:

(

) (

) , (90)

using a skin drag coefficient as 0.008 and end drag coefficient as 0.08 Each BCM, since they are

cylindrical in shape use OrcaFlex’s default properties for axial added mass and normal added

mass .

50

4 NUMERICAL RESULT

This chapter presents numerical estimates of OCT performance calculated using the

developed numerical modeling tools. Section 4.1 presents results calculated using the developed rotor

model alone, and compares these performance predictions with those calculated using published rotor

models. Section 4.2 presents numerical results calculated using the developed complete OCT

numerical system model attached to a mooring. These results highlight the utility of the numerical

model and provide predictions for the performance of a moored OCT.

4.1 Model Validation

This section presents the numerical validation of the developed BEM rotor model. The

selected operating scenarios are utilized to quantify the developed rotor model performance estimates

for steady state operations (4.1.1), the response of the momentum model to transient events (4.1.2) and

a wave environment (4.1.3). For these comparisons the developed numerical rotor model and the two

models utilized for validation, WT_Perf [38] (steady operations only) and Matlab [9] each used 100

elements for each blade. It is important to note that these in these comparisons Prandtl’s hub loss and

advanced brake-state models were disabled for WT_Perf, since these models were not utilized in the

other two numerical models.

4.1.1 Steady State

The steady operating condition analysis is the first stage of the validation process. In this

stage the coefficients of shaft power, torque and rotor thrust (drag) are analyzed, along with the shaft

power and restoring moment. These coefficients are calculated using the following non-dimensional

equations:

51

(91)

(92)

(93)

where is the coefficient of power, is the shaft power, is the density of water, is the swept

rotor plane area, is the free-stream velocity magnitude, is the coefficient of torque, is the shaft

torque, is the radius of the rotor blade, is the coefficient of thrust and is the thrust.

Two steady flow operating conditions are evaluated in this sub-section, which are utilized to

validate the developed rotor model against previously published numerical models. The first of these

conditions is a steady axial flow impacting a rotor rotating at a constant RPM. These results are

compared with the predicted steady state axial flow rotor performance calculated using the wind

turbine code WT_Perf [38], as well as a previously mentioned Matlab model [9] (sub-section 4.1.1.1).

The second operating condition quantifies the behavior of the OCT rotor for a steady but non-axial

flow. This off axis flow analysis compares the results for , power and the restoring moments about

the -axis when a progressive yaw is applied to the OCT of up to 30°, with results calculated using the

Matlab model.

4.1.1.1 Axial Flow

Axial flow testing allows for the prediction of the maximum shaft power output of the OCT,

as well as the relationship between TSR and power, torque, and thrust; which are important for

selecting the proper generator and gear box for maximum energy production, conducting heat

dissipation analyses, and estimating the potential energy production from OCTs. The coefficients of

power, torque and thrust are plotted against TSR in order to define the performance over the projected

operating range of tip speed ratios (Figure 12). In this figure the green lines represent Matlab based

52

predictions, black lines represent WT_Perf based predications, and red lines represent the results from

the developed numerical model. The data produced using the developed numerical model is also

presented in Table 4.

Figure 12: The predicted coefficients of power, thrust (drag) and torque using Matlab Model (green), WT_Perf

model (WTP)(black) and the OrcaFlex model (OF)(red *) as listed in the legend.

These results show that the predicted performance coefficients from the three numerical

models are very similar, especially at TSRs where maximum power is produced (from around 2-6).

The results for the maximum values calculated by the model developed in this thesis are ,

and occuring at the TSRs of 4.10, 3.50 and 2.00, respectively. The Matlab

model had maximums that were calculated to be at TSRs of 3.97, 3.65 and 2.05 which yield

, and , respectively. The WT_Perf model produced similar results with

53

maximum values of , and at the corresponding TSRs of 4.05,

3.65 and 2.08.

Table 5: This table shows the values for , and at a flow speed of 1.6 m⁄s calculated using the developed

rotor model.

4.1.1.2 Off axis flow

The off axis steady and homogeneous flow analysis characterizes OCT performance for the

likely scenario where the OCT is not precisely aligned with the oncoming flow. These time varying

forces that cyclically impact the blades each rotation can lead to fatigue failures or excite natural

frequencies of the system. For this evaluation the numerical predictions were made using an

undisturbed flow speed of 1.6 m/s. A rotor RPM of 41.74 was used in this analysis, resulting in a TSR

of 4.10 (corresponding to the maximum CP for an axial flow).

The relationship between mean shaft power and with respect to the off-axis angle is

presented in Figure 13.a and the relationship between restoring moment and off-axis angle is given in

TSR RPM TSR RPM

0.00 0.0000 0.0000 0.0303 -0.1975 4.50 45.8366 0.4361 0.0944 -0.6649

0.25 2.5465 0.0092 0.0360 -0.1969 4.75 48.3831 0.4312 0.0884 -0.6580

0.50 5.0930 0.0237 0.0461 -0.2064 5.00 50.9296 0.4243 0.0827 -0.6489

0.75 7.6394 0.0469 0.0609 -0.2294 5.25 53.4761 0.4158 0.0771 -0.6375

1.00 10.1859 0.0834 0.0812 -0.2712 5.50 56.0225 0.4052 0.0718 -0.6241

1.25 12.7324 0.1374 0.1070 -0.3390 5.75 58.5690 0.3925 0.0665 -0.6086

1.50 15.2789 0.2078 0.1349 -0.4296 6.00 61.1155 0.3775 0.0613 -0.5911

1.75 17.8254 0.2815 0.1567 -0.5215 6.25 63.6620 0.3600 0.0561 -0.5717

2.00 20.3718 0.3384 0.1648 -0.5865 6.50 66.2085 0.3398 0.0509 -0.5502

2.25 22.9183 0.3728 0.1614 -0.6101 6.75 68.7549 0.3167 0.0457 -0.5268

2.50 25.4648 0.3959 0.1542 -0.6310 7.00 71.3014 0.2908 0.0405 -0.5014

2.75 28.0113 0.4114 0.1457 -0.6479 7.25 73.8479 0.2618 0.0352 -0.4741

3.00 30.5577 0.4220 0.1370 -0.6601 7.50 76.3944 0.2295 0.0298 -0.4448

3.25 33.1042 0.4292 0.1286 -0.6682 7.75 78.9409 0.1938 0.0244 -0.4137

3.50 35.6507 0.4342 0.1208 -0.6730 8.00 81.4873 0.1546 0.0188 -0.3805

3.75 38.1972 0.4374 0.1136 -0.6747 8.25 84.0338 0.1118 0.0132 -0.3455

4.00 40.7437 0.4387 0.1068 -0.6733 8.50 86.5803 0.0652 0.0075 -0.3085

4.10 41.7623 0.4388 0.1042 -0.6722 8.75 89.1268 0.0147 0.0016 -0.2696

4.25 43.2901 0.4385 0.1005 -0.6700 9.00 91.6732 -0.0398 -0.0043 -0.2288

54

Figure 13.b. Figure 13.a shows that as the off-axis angle increases away from the flow direction a

significant reduction in is similarly predicted by both models. The specific amount of shaft power

lost at 30° yaw from the flow, in the OrcaFlex model, is slightly higher than 21% compared to the

23% loss calculated in the Matlab model. The restoring moments calculated using the two models

increase to approximately 240 N-m for an off axis angle of 30°C. Over this range the greatest

difference occurred at 15°, with a value of 17.20 N-m.

Figure 13: Predicted coefficient of power (top left), power (top right) and compared restoring moment

(bottom)(Matlab model black line OrcaFlex model blue line) from a 3 m rotor diameter operating at a TSR of

4.10. These results are presented as a function of off axis flow angle.

4.1.2 Transient State

The analyses in this section compare the calculated transient response of the OrcaFlex and

Matlab BEM rotor models to validate the utilized unsteady momentum model. This is done by

independently comparing the numerically estimated transient rotor performance when subject to step

changes in RPM and flow speed. During the first of these tests, the RPM is initially 41.76 (the RPM

where maximum shaft power is generated) until the rotor performance converges to nearly a steady

value. After 9 seconds the RPM is abruptly decreased to 20.37 and the simulation runs until the rotor

model once again approaches steady state operation (Figure 14).

55

Figure 14: Predicted power response when RPM is changed from 41.76 (TSR 4.10) to 20.37 (TSR 2.0)

Figure 14 shows the rotor codes response to abrupt changes in RPM, with a constant flow

speed, and flow speed, with a constant RPM, respectively as well as show the rate at which they

converge. The -axis are limited to better show the response to the step changes but for completeness

it is worth noting that the initial shaft power is approximately and this is due to both

numerical models assuming that initially there is no reduction on flow speed caused by the rotor

forces. These results show that it initially took approximately 2.35 s and 2.60 s for the OrcaFlex and

Matlab, respectively, to converge within 1% of their first steady state value of 6.87 kW for OrcaFlex

and a value of 6.98 kW for the Matlab model. After the step change in RPM it takes approximately

1.05 s for the OrcaFlex model to come within 1% of its steady state value of 5.29 kW, while the

Matlab model took 0.75 s to come within 1% of its steady state value of 5.33 kW.

56

The second test evaluates the response to a step change in the flow magnitude from 1.6 m/s to

1.4 m/s. The results collected from the developed OrcaFlex simulation tool is compared to similar

tests that were performed by the Matlab simulation in Figure 15. The same analysis was done for the

current shift simulation and produced the identical 9 s convergence data. For the step change in

current speed from 1.6 m/s to 1.4 m/s it took 1.6 s for OrcaFlex to come within 1% of its steady state

value of 4.34 kW while Matlab took 1.8 s to come within 1% of its steady state value of 4.57 kW.

Figure 15: Predicted power response when flow speed is changed from 1.6 m/s to 1.4 m/s

Drastic step changes, for RPM and current speed, were used to test the convergence rate of

the models rotor power response, it also showed that at different RPMs and current speeds the

OrcaFlex model was still comparable in terms of the accuracy of the data.

57

4.1.3 Waves

This phase of the validation is used to quantify the impact waves have on calculated rotor

performance. This test evaluates rotor performance for the developed numerical model using a

JONSWAP wave spectrum with a 2 m significant wave height, a peak period of 7 s, 40 wave

components for each propagation direction, waves propagating from 10 directions, and a wave

direction exponential spreading exponent of 24. The Matlab on the other hand uses a Pierson-

Moskowitz Spectrum with the same values. The two different wave fields were used due to the Matlab

model being written to model Pierson-Moskowitz wave spectrum and which is not offered in the

OrcaFlex environment. The main difference between these two spectra is that the Pierson-Moskowitz

assumes the concept of a fully developed sea while the JONSWAP assumes the wave spectrum never

develops and continues to evolve through non-linear wave-wave interactions [39].

For this analysis the depth of 20 m below the mean sea surface is considered for both

numerical models. The rotor shaft is assumed to maintain a fixed location, pitch, and heading while

the rotor rotates at a fixed RPM. When looking these data it is important to note that different wave

spectra and initial phase angles for the individual wave components were used and therefor the time

histories will be uncorrelated and the wave fields will be statistically similar, but not identical. Other

contributing factors to discrepancies in forces include the developed OrcaFlex numerical model

assuming that the undisturbed free stream velocity is constant over the entire swept area of the rotor

and only varies with time. For the analysis presented in this section, numerical simulations were run

for wave fields with mean propagation directions in the same direction as the free stream flow velocity

(axial with respect to the rotor), which is aligned with the rotor axis, and for waves that are

perpendicular to the flow speed (perpendicular to the rotor).

For axial waves the calculated shaft power (Figure 16) is predicted to have a greater wave

induced variance by the OrcaFlex model, with a standard deviation of 1.595 kN (22.7% of the mean),

when compared to the Matlab model, with a standard deviation of 0.543 kN (7.8% of the mean). This

is likely due to the application of force to one central location on the rotor and not being dispersed

58

across the swept rotor plane area. The mean shaft power calculated using these numerical models

differs by less than 1%. Because of the fundamental differences between the two models, primarily

driven by the OrcaFlex interface limitations, these results suggest that it is important to consider the

impact of these modeling limitations when evaluating turbine performance predictions if waves are

one of the dominant environmental stimuli.

Figure 16: Shaft Power for a constant RPM of 41.76 experiencing waves

This analysis also compared the net forces on the rotor in the body fixed , and directions

(Figure 17). This figure shows that the mean axial force is increased by the waves and that the time

varying vertical and lateral forces are also created by the waves. The wave induced moments

presented in Figure 18 are the net moments experienced by the rotors shaft. There was a small percent

difference present for the mean value of the axial forces with OrcaFlex predicted values being 1.49%

higher with a value of 6.37 kN compared with 6.73 kN from the Matlab model. However, standard

deviation is much greater for the OrcaFlex model, 1.037 kN, than Matlab model’s value of 0.342 kN.

A full list of standard deviations and variances are presented in Table 5. The moments in Figure 18

showed less deviation than the forces yet not significantly difference in terms of percent difference.

The difference in standard deviation for the moments between the OrcaFlex and Matlab model was

239.49 N-m (363.70 N-m for OrcaFlex and 124.21 N-m for Matlab), 4.82 N-m (21.28 N-m and 16.46

59

N-m) and 8.73 N-m (15.58 N-m and 6.85 N-m) for the , and directions, respectively with the

largest standard deviations occurring in power and the direction force and moment.

Figure 17: Forces experienced by a rotor shaft with a fixed RPM of 41.76 when subjected to waves

Figure 18: Shaft moments experienced by a rotor with a fixed RPM of 41.76 when subjected to waves

60

Table 6: Comparisons of standard deviations and variances of the turbine code when subjected to axial waves

and 90° off axis waves

The off-axis waves presents a very different view for which model experienced the greatest

simulated wave impacts. This analysis shows the greater impact of the wave is felt by the Matlab

model than the OrcaFlex model. As expected the off-axis waves appear to have a significantly lower

cyclic impact on the rotors predicted power curve at a depth of 20 m (Figure 19). However, like the

previous wave data the percent difference in the mean shaft power was less than 1% as it only varied

by 51.2 W. Even though the means were similar in the axial and off axis wave environment the

difference in standard deviation was much lower, 90.84 N, yielding a much smaller percent difference

value, 18.39% then then axial simulation.

Figure 19: Shaft power for a constant RPM of 41.76 operating at a depth of 20 m with incoming waves that are

90° off axis.

OrcaFlex Matlab OrcaFlex Matlab OrcaFlex Matlab OrcaFlex Matlab

1,594.50 543.22 Power(W) 448.55 539.39 2,542,300.00 295,090.00 Power(W) 210,200.00 290,940.00

1,037.20 341.74 X 270.07 338.95 1,075,900.00 116,790.00 X 72,935.00 114,890.00

10.72 6.11 Y 32.13 5.88 114.87 37.31 Y 1,032.50 34.56

33.25 10.94 Z 32.24 10.82 1,105.70 119.58 Z 1,039.50 116.97

363.70 124.21 X 102.32 123.34 132,280.00 15,428.00 X 10,468.00 15,215.00

21.28 16.46 Y 24.58 16.22 452.81 271.05 Y 604.21 263.01

15.58 6.85 Z 23.14 6.86 242.67 46.93 Z 535.62 47.10

Variance

Forces (N)

Moments (N-m)

Axial Waves 90° Off Axis Waves

Moments (N-m)

Forces (N)

90° Off Axis Waves

Standard Deviation

Axial Waves

61

The values for the forces and moments experience by the shaft in off axis waves also vary in a

similar pattern to the axial waves. The differences in the standard deviation for the forces experienced

by the shaft between the OrcaFlex and Matlab model was on 68.88 N (270.07 N for OrcaFlex and

338.95 N for Matlab), 26.25 N (32.15 N and 5.88 N)and 21.42 N (32.24 N and 10.82 N) in the body

fixed frame , and directions respectively while the differences of the standard deviation for the

moments experienced by the shaft for the two models varied 21.02 N-m (102.32 N-m for OrcaFlex

and 123.34 N-m), 8.36 N-m (24.58 N-m and 16.22 N-m) and 16.28 N-m (23.14 N-m and 6.86 N-m)

for the , and -directions respectively as seen in Table 5.

Figure 20: Forces applied to the rotor for a constant RPM of 41.76 operating at a depth of 20 m with incoming

waves that are 90° off axis.

Figure 21: Moments experienced by the rotor for a constant RPM of 41.76 operating at a depth of 20 m with

incoming waves that are 90° off axis.

62

4.2 Moored Turbine Performance

This section presents the numerical results from the simulation of the OCT attached to a

single cable and operating in a steady current. This system is depicted in Figure 22, with one end of

the mooring ling fixed in space and the other attached to the OCT. The steady current that are being

applied is the mean current value of 1.6 m/s measured for a depth of 20 meters during a 13 month

study (from February 2009 – March 2010) at a potential mooring location for OCT [9]. This section

addresses the performance of the OCT for the mean current.

Figure 22: Mooring configuration utilized to simulate OCT performance in a uniform steady current.

Figure 23: This figure shows the power versus time for the rotor for a steady 1.6 m/s current

Figure 23 shows the expected power predictions for the turbine are close to what is expected

from it settling out to about 7.1 kW during this simulation and reaching a maximum, 7.8 kW after the

momentum model converged.

63

During this simulation the OCT has a small rotation about the -axis (roll) (Figure 24). This

roll has a maximum amplitude of 2.66° (at 9.9s) and due to the design of SNMREC’s OCT; where the

hydrostatic moment is used to balance the rotor torque.

Figure 24:This figure shows the rotation of the OCT about the x-axis (roll) for a steady 1.6 m/s current with a

constant rotor rotational velocity 41.76 RPM.

The effective line tension of the mooring line that is attached to the OCT at one end and

anchored to a fixed location at the other end is presented in Figure 25. The maximum tension

experienced by the line was 1.118 kN (at 5.5 s), which occurs at nearly the same time as the maximum

distance between the turbine and the geodetically fixed anchored end of the mooring.

Figure 25: This figure shows the tension in the mooring line attached to the OCT in relation to time for steady

1.6m/s current with a constant rotor rotational velocity 41.76 RPM.

The simulation presented in this section are significantly shorter than the validation

simulations due to increased variable size and memory limitations with increased computational space

requirements required for multiple file computations. These limitations led short simulation time

64

which did not allow the system to reach steady state. The type of variables used and large array size

leads to a heap memory, a large pool of memory used for dynamic allocation in C++, error that will

not allow the .dll file to be created. By reducing the number of variables being saved during the

simulation, the length of the simulation can be extended. With a system with more allocated dynamic

memory, the created file can be as large as the operating system format allows and would allow for the

system to reach steady state.

65

5 CONCLUSIONS AND SUGGESTED FUTURE WORK

5.1 Conclusions

A numerical modeling tool for OCT was developed utilizing the OrcaFlex dynamic marine

modeling software. This tool utilizes basic OrcaFlex elements and custom developed external

functions that enable the utilization of BEM rotor theory and account for the additional degree of

freedom associated with the rotating rotor. The developed external function based rotor model was

validated independent from the rest of the system. Upon completion of the validation, the forces

generated by rotor model were applied to a single line moored buoy to evaluate the response of the

system. Finally, the rotor model was applied to SNMREC’s experimental turbine system which

included the OCT an MTB and an ODCP as well as its full mooring system.

The numerical model of the rotor is a native external function that was developed in the

coding language C++ to allow continuous access to OrcaFlex variables and incorporates rotor forces

back into the simulation each time step. This external function allows the user to set the rotor RPM,

number of blades, the number of elements and the number of mesh points use in the rotor calculations.

This external function also utilized OrcaFlex defined environmental data and system states.

Unfortunately, due to software limitations in the OrcaFlex environment, the instantaneous water

velocity used in these calculations is limited to a single fluid velocity vector calculated for the buoy’s

wetted centroid and not the entire swept area. This external function assumes not fatigue or

catastrophic failures and is currently set to simulate SNMREC’s experimental 20 kW. This model

predicted the optimal tip speed ratio to be 4.10, equating to an RPM of 41.76 for a 1.6 m/s uniform

current velocity. When compared to two other published models the coefficient of power, lift and drag

were all below 2% at the optimal tip speed ratio.

66

The performed simulations of the geodetically fixed turbine rotor were then subjected to step

changes in the RPM values and current velocity values to validate and compare the rotor response and

convergence rate. The response in shaft power to a step change in RPM took 28.6% more time than

the compared model to converge to within 1% of the steady state shaft power of 5.29 kW, taking 1.05

s. The predicted response to the step RPM change showed less than 2% difference in the steady state

power between the two models. The shaft power response to the step water speed change predicted

using the OrcaFlex tool took 12.5% less time to converge to within 1% of its steady state value of 4.34

kW, taking 1.6 s. These results show that the moment model utilized in the tool reacts at a very similar

rate to existing models, which is an important model parameter for evaluating system performance in

a wave field.

The evaluation of the developed rotor model under wave excitation, highlights the impact of

one of the limitations of the utilized approach. The OrcaFlex external function option only provides

the user with relative water velocity at the center of volume of the body that calls this function.

Therefore, unless a process is developed to work around this limitation, external functions will be

limited to assuming that the water velocity is homogeneous over the swept area of the rotor blade (the

assumption utilized in this thesis). It is assumed that this modeling assumption is primarily responsible

for the discrepancies in the variability of the calculated power outputs and forces from the two models.

For the simulation generating axial wave excitation the standard deviation of the power generated for

the OrcaFlex tool was 1.595 kN and the model for comparison yielded a standard deviation of the

power generated to be 0.543 kN. For the forces and moments calculated for an axial wave field the

OrcaFlex model predicted standard deviation values more than three times greater than the other

model.

In the full systems testing consisting of the rotor code, the forward and rear nacelle, the two

BCMs and a mooring line, subjected the system to a uniform current profile of 1.6 m/s. This

simulation demonstrates a convergent behavior, though steady state is not reached for the full system

simulations. Due to the time steps used in the simulation as well as length of the simulations, steady

67

state cannot be reached; this is due to memory management in the external function code as well as

limited dynamic memory available in developing dynamic loading library file which is needed to

communicate with OrcaFlex.

5.2 Suggested Future Work

For the first step in the future work should be to do a complete OCT analysis using the

intended mooring system in the simulation.

Suggested improvements to the developed numerical simulation tool include overcoming the

OrcaFlex external function limitation that only provides water velocities at one location to external

functions. This will likely require direct collaboration with OrcaFlex engineers to either access an

array of water velocities each time step, or by directly accessing the algorithms that OrcaFlex uses to

create its wake field. This will improve the ability of the developed code to predict performance when

current shear exists or when operating in waves.

Another possibility for future work includes increasing the usability of the tool by optimizing

data input methods from other sources, as well as improving the memory management within the tool.

The majority of SNMREC’s rotor data is hard coded into the modeling tool requiring a full reworking

of several of the header files that make up the project. The development of a routine that reads device

data directly from an excel sheet would greatly increase the value of this tool for future applications.

68

REFERENCES

[1]F.L. Ponta, P.M. Jacovkis, Marine-current power generation by diffuser-augmented floating hydro-

turbines, Renewable Energy, Volume 33, Issue 4, April 2008, Pages 665-673

[2]Takahashi P, TrenkaA. Ocean Thermal Energy Conversion. New York: Wiley, 1996.

[3]M.S. Güney, K. Kaygusuz, Hydrokinetic energy conversion systems: A technology status review,

Renewable and Sustainable Energy Reviews, Volume 14, Issue 9, December 2010

[4]Ben Elghali, S.E.; Benbouzid, M.E.H.; Charpentier, J.F.; , "Marine Tidal Current Electric Power

GenerationTechnology: State of the Art and Current Status," Electric Machines & Drives

Conference, 2007. IEMDC'07. IEEE International , vol.2, no., pp.1407-1412, 3-5 May

2007

[5]Mukrimin, Sevket, Guney, Evaluation and measures to increase performance coefficient of

hydrokinetic turbines, Renewable and Sustainable Energy Reviews, Volume 15, Issue 8,

October 2011, Pages 3669-3675, ISSN 1364-0321, 10.1016/j.rser.2011.07.009.

[6]Fergal0Rourke,FergalBoyle,AnthonyReynolds,Tidalenergyupdate009,

AppliedEnergy,Volume87,Issue2,February2010,Pages398-409,ISSN0306-

2619,10.1016/j.apenergy.2009.08.014.

[7]HYCOM, http://hycom.org/dataserver/

[8] J.H. VanZwieten Jr., A.E.S. Duerr, G.M. Alsenas, and H.P. Hanson (2013) “Global ocean current

energy assessment: an initial look” Proceedings of the 1st Marine Energy Technology

Symposium (METS13) hosted by the 6th

annual Global Marine Renewable Energy

Conference, April 10-11, Washington D.C. Available:

http://www.foroceanenergy.org/mets/2013-peer-reviewed-mets-papers/

[9]VanZwieten, J.H., Vanrietvelde, N., and Hacker, B., 2012, “Numerical simulation of an

experimental ocean current turbine” Oceanic Engineering, In-Press, no. 2010JOE000780

[10]David L.F. Gaden, Eric L. Bibeau, A numerical investigation into the effect of diffusers on the

performance of hydro kinetic turbines using a validated momentum source turbine model,

Renewable Energy, Volume 35, Issue 6, June 2010, Pages 1152-1158, ISSN 0960-1481,

10.1016/j.renene.2009.11.023.

[11]A Discrete Vortex Method for Simulating a Stand-Alone Tidal-Current Turbine: Modeling and

Validation Ye Li and Sander M. Calışal, J. Offshore Mech. Arct. Eng.

132, 031102 (2010), DOI:10.1115/1.4000499

[12]Fergal O. Rourke, Fergal Boyle, Anthony Reynolds, Renewable energy resources and

technologies applicable to Ireland, Renewable and Sustainable Energy Reviews, Volume

13, Issue 8, October 2009, Pages 1975-1984, ISSN 1364-0321, 10.1016/j.rser.2009.01.014.

[13]Marine Current Turbines press release, 18th March 2004.

[14]Myers, L. and Bahaj, A. S. Basic operational parameters of a horizontal axis marine current

turbine.In Proceedings of 8th World Renewable Energy Congress, Denver, USA, 2004.

[15]Pham, X., Liu, Y., Varyani, K., and Barltrop, N. Report on tidal stream rotor tests. Universities of

Glasgow and Strathclyde, 2004.

[16]VanZwieten, J. 2003. Modeling and control of the “C-Plane” ocean current turbine.Masters

Thesis, Florida Atlantic University, or VanZwieten, J., Driscoll, F.R., Leonessa, A., Deane,

G., 2005.Design of an ocean current turbine. Part I: Mathematical modeling and dynamics

simulation. Ocean Engineering, submitted for publication.

[17]MOL Hansen “Aerodynamics of wind turbines”Earthscan, London, UK. 2008 Chapter 6

69

[18]Vanrietvelde, N. NUMERICAL PERFORMANCE PREDICTION FOR FAU'SFIRST

GENERATION OCEAN CURRENT TURBINE, Florida Atlantic University. 2009

[19]Cribbs, A.R.; VanZwieten, J.H.; , "Global numeric analysis of a moored ocean current turbine

testing platform," OCEANS 2010 , vol., no., pp.1-7, 20-23 Sept. 2010 doi:

10.1109/OCEANS.2010.5664288

[20]Cribbs, A.R. 2010, Model analysis of a mooring system for an ocean current turbine testing

platform, Florida Atlantic University.

[21]Young, M. T.;Design and Analysis of an Ocean Current Turbine Performance Assessment

System, Florida Atlantic University

[22]He, C. 1989.Development and Application of a Generalized Dynamic Wake Theory for Lifting

Rotors. Atlanta: Georgia Institute of Technology, July

[23]J.F. Manwell, J.G. McGowan, A.L. Rogers, Wind Energy Explained:Theory, Design and

Application, John Wiley & Sons Ltd., West Sussex, Britain, 2002 (reprinted with

corrections Aug 2003)

[24]Mason W.H., Applied Computational Aerodynamics – Lecture Notes, Department of Aerospace

and Ocean Engineering, Virginia Polytechnic Institute and State University, Volume 1,

Chapter 6, 1998. http://www.aoe.vt.edu/~mason/Mason_f/CAtxtTop.html

[25]Goly, A.; Ananthakrishnan, P.; "Hydrodynamic analysis of a gulf-stream turbine using the vortex

lattice method,"OCEANS 2010, vol., no., pp.1-10, 20-23 Sept. 2010doi:

10.1109/OCEANS.2010.5664089

URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5664089&isnumber=566

3781

[26]Madsen, H. A. and Rasmussen, F. A near wake model for trailing vorticity compared with the

blade element momentum theory. Wind Energ., 2004, 7:325–341. doi:10.1002/we.131

[27]Glauert H. Airplane propellers. In Aerodynamic Theory, Durand WF (ed.). Dover: New York,

1963; 169–360.

[28]Snel, H. Review of the present status of rotor aerodynamics. Wind Energ. (1998)

[29]Mustafa Cavcar, Blade Element Theory—Lecture Notes,School of Civil Aviation,Anadolu

University 2004, Eskisehir, Turkey.

URL:http://home.anadolu.edu.tr/~mcavcar/hyo403/Blade_Element_Theory.pdf

[30]Tip Loss Correction for Actuator/Navier--Stokes Computations Wen ZhongShen, Jens Norkaer

Sorensen, and Robert Mikkelsen, J. Sol. Energy Eng. 127, 209 ,2005

[31]Prandtl L, Betz A. VierAbhandlungenzurHydrodynamik und Aerodynamik. GöttingerNachr.:Göttingen, 1927; 88–92

[32]Shen, W. Z., Mikkelsen, R., Sørensen, J. N. and Bak, C., Tip loss Corrections for Wind Turbine

Computations. Wind Energ. 2005, 8: 457–475.

[33]R. Lanzafame, M. Messina, Fluid dynamics wind turbine design: Critical analysis, optimization

and application of BEM theory, Renewable Energy, Volume 32, Issue 14, November 2007,

Pages 2291-2305, ISSN 0960-1481, 10.1016/j.renene.2006.12.010.

[34]Orcina Ltd. http://www.orcina.com/

[35]OrcaFlex Theory Manual.

http://www.orcina.com/SoftwareProducts/OrcaFlex/Documentation/OrcaFlex_US_paper_s

ize.pdf

[36]T.I. Fossen. Guidance and Control of Ocean Vehicles. Wiley, Chichester, U.K., 1994.

[37]Narayana Mahinsasa . Catastrophic Failure due to Gyroscopic Effect of Small Scale Tilt up

Horizontal Axis Wind Turbines, DEWI Magazin Nr. 15, August 1999.

[38]M. L. Buhl, “WT_PERF user’s guide,” National Wind Technology Center – National Renewable

Energy Laboratory, Golden, CO, 2004,

[39]Robert Stewart. Introduction to Physical Oceanography, 2008 URL

http://oceanworld.tamu.edu/resources/ocng_textbook/chapter16/chapter16_04.htm#top


Recommended