• No results found

Experimental validation of a lumped-mass mooring line model with scaled mussel line test data

N/A
N/A
Protected

Academic year: 2021

Share "Experimental validation of a lumped-mass mooring line model with scaled mussel line test data"

Copied!
94
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

line model with scaled mussel line test data

Experimental validation of a lumped-mass mooring

Academic year 2019-2020

Master of Science in Electromechanical Engineering

Master's dissertation submitted in order to obtain the academic degree of Supervisors: Prof. dr. ir. Evert Lataire, Ajie Brama Krishna Pribadi Student number: 01404822

(2)
(3)

line model with scaled mussel line test data

Experimental validation of a lumped-mass mooring

Academic year 2019-2020

Master of Science in Electromechanical Engineering

Master's dissertation submitted in order to obtain the academic degree of Supervisors: Prof. dr. ir. Evert Lataire, Ajie Brama Krishna Pribadi Student number: 01404822

(4)

Acknowledgements

This master’s thesis is the final step of my academic career at Ghent University. Writing it would not have been possible without the support of several people.

Firstly, I would like to thank some members of the Maritime Technology Division at Ghent University. I thank Prof. dr. ir. Evert Lataire for introducing me to the subject and being available at any time. Next, I thank ir. Luca Donatini and ir. Ajie Pribadi for helping me get to know the MoorDyn-UGent software and resolving countless issues.

I would also like to express my gratitude to NTNU and its staff for giving me the opportunity to execute my experiments in the MC-Lab in Trondheim. In particular, I thank Prof. Pål Furset Lader for his help in setting up the experiments and for sharing his experience in the field of model testing. I also thank ir. Emil Bratlie and ir. Torgeir Wahl for assisting me during the experiments and providing me with adequate equipment.

Last but not least, I would like to thank my family for their support during the realisation of this dissertation and throughout my entire studies.

Roeland Oppeel Ghent, May 2020

The author gives permission to make this master dissertation available for consultation and to copy parts of this master dissertation for personal use. In the case of any other use, the copyright terms have to be respected, in particular with regard to the obligation to state expressly the source when quoting results from this master dissertation.

(5)

Experimental validation of a lumped-mass mooring line model

with scaled mussel line test data

Roeland Oppeel

Supervisors: Prof. dr. ir. Evert Lataire, ir. Ajie Pribadi, ir. Pal Lader

Abstract

The present study is an experimental validation of the numerical mooring dynamics model MoorDyn-UGent. Regular waves of altering wave period, wave height and orientation were generated and applied to a 1:20 scaled model of a semi-submerged mussel longline. Tension forces in the mooring chains and the position of the buoys were recorded. The same exper-iments were simulated in the numerical model. The simulated and measured outcomes were then compared.

In the introductory chapter 1, aquaculture and the possible co-use of area are described. The origin of the MoorDyn-UGent numerical model within the scope of the Edulis project is ex-plained. The theoretical principles of this numerical model are given in chapter 2. First, the original open-source MoorDyn by Matthew Hall is looked at. The adaptations made by the Maritime Technology Division of Ghent University are mentioned afterwards. Chapter 3 de-scribes the methodologies used within this project. The physical experiments are explained, as well as how they are translated into the numerical model. The results are summarized in chap-ter 4, along with a short evaluation. Finally, conclusions and considerations for future work are given in chapter 5.

Keywords

(6)

Experimental validation of a lumped-mass mooring line model with

scaled mussel line test data

Roeland Oppeel

Supervisors: Prof. dr. ir. Evert Lataire, ir. Ajie Pribadi, ir. Pal Lader Abstract – Experiments with regular waves have

been performed in order to validate the numerical model MoorDyn-UGent. A scaled model of a mussel longline was built and subject to waves with altering wave height and period. The forces in the mooring chains were measured, as were the positions of the buoys. The same experiments were modelled in the MoorDyn-UGent software. Measured and simulated results were compared in order to validate the numer-ical model. This model could eventually be used to de-sign the optimal configuration for mussel farms in off-shore areas.

Keywords –Aquaculture, numerical modelling, ex-perimental validation

I. Introduction

Aquaculture is generally seen as the sector with the greatest potential to answer the growing demand for food [1]. Prime candidates for aquaculture are blue mussels (Mytilus Edulis) due to their ability to withstand wide fluctuations in salinity, temperature and oxygen levels. The finite area of protected near-shore waters and the high number of stakeholders limit the development of coastal aquaculture. Cultivating mus-sels further offshore not only reduces the amount of possible conflicts, but leads to a better growth rate and shorter pro-duction cycles [2]. Several studies have assessed the biolog-ical, technical and economic expediency of offshore mussel aquaculture, showing good results [3, 4].

Of the techniques commonly practised, longlines are most easily adapted to utilisation in harsh weather condi-tions [5]. A longline consists of a horizontal line called the backbone which is moored with chains and anchors. Buoys along the line provide the buoyancy. Ropes suspended from the backbone act as substrates for the species to grow on. The longline can be floating, semi-submerged or submerged. Submerging the lines reduces the effect of waves and thus the motion and loads on the lines.

The Belgian part of the North Sea contains many stake-holders. One possibility to lower the chances of user con-flicts is the co-use of area. The combination of offshore wind farms and mussel aquaculture stands out with the great-est potential. Studies by Buck et al. and Lagerveld have acknowledged the feasibility of bringing the two together [6, 7]. A well-thought regulatory framework is however needed to initiate the combination. In Belgium, the Edulis project has been initiated. This project unites research in-stitutes and industrial partners: ILVO, Ghent University, OD Natuur, C-Power, Belwind, DEME, Sioen Industries, Colruyt Group, Brevisco and Lobster Fish. It assesses the technical and ecological feasibility of cultivating blue mussels in the

Belgian wind farms. Two test lines have already been de-ployed: a Bio line in the C-power concession zone and a Force line in the Belwind wind farm. The faculty of Bio-science Engineering of Ghent University assesses the eco-logical feasibility. The Maritime Technology Division (MTD) of the Faculty of Engineering and Architecture deals with the technical aspects. These include modelling the forces and motions of the system and designing an optimal lay-out of a mussel farm. The MTD developed the numerical program MoorDyn-UGent to this end. This software pre-dicts the behaviour of moored systems under environmental loads.

This research project concerns an experimental valida-tion of the MoorDyn-UGent software. A 1:20 scaled model of a semi-submerged longline (the Edulis Force line) is sub-ject to a range of incoming waves. The wave height, period and incoming wave angle are altered. Buoy positions and tension forces in the chains are measured and compared to simulated values by the numerical model.

II. Theory

MoorDyn-UGent is an adapted version of the open-source code MoorDyn. MoorDyn is a lumped-mass mooring line model developed by Matthew Hall [8]. It is used to calculate the dynamics of mooring systems. To this end, the motion of fairleads is needed, often generated by a coupled program. The MTD modified the software in order to include environ-mental inputs.

A. MoorDyn

In MoorDyn every line is divided into N evenly-sized seg-ments and N+1 nodes. The mass of every segment is trans-ferred to the adjacent nodes. The lines are assigned a length, diameter, density and Young’s modulus. Bending and tor-sional stiffness are neglected. Buoys and anchors are given a mass and a volume in the input file. Internal and exter-nal forces are calculated at the nodes. Damping forces, axial stiffness and net weight make up the internal forces. Exter-nal forces comprise hydrodynamic forces according to Mori-son’s equation and forces originating from contact with the bottom. The original MoorDyn code does not account for wave kinematics.

When MoorDyn is run, an initial equilibrium state is de-termined. The mooring system is allowed to settle simu-lating a dynamic relaxation with boosted drag coefficients. Next, MoorDyn calculates the forces and accelerations of all the nodes at every time step. The user defines what proper-ties are printed to the output.

(7)

B. MoorDyn-UGent

The working principle of the adapted version of MoorDyn is much the same as for the original. The most important adaptation concerns the inclusion of environmental loads i.e. waves and current. The current profile is described in the main input file. A separate file contains the properties of the waves. Irregular waves can be generated by entering all components in the input file. Another addition is the pos-sibility to assign width and height to a buoy. This way the hydrodynamic behaviour of the buoys is modelled. Buoys are assumed cylindrical and always upright.

III. Methodology

Regular waves were applied to a scale model of a semi-submerged longline. Wave heights ranged from 10 to 30 cm with a step of 5 cm. Wave periods were increased from 0.5 s to 2.5 s with a step of 0.5 s. A wave gauge was installed to capture the properties of the physical waves. Four different incoming wave angles were as-sessed.0° coincides with waves parallel to the longline. The tests were performed at the Marine Cybernetics Laboratory (MC-Lab) at NTNU (https://www.ntnu.edu/imt/lab/ cybernetics). This lab is a small wave basin with tank dimensions L x B x D =40 m x 6.45 m x 1.5 m. The x-axis of the right-handed inertial reference system lies along the length of the tank with the same direction as the waves. The z-coordinates are positive upwards. This system will be used throughout this project. The buoy positions were measured using Qualisys Oqus cameras and the Qualisys Track Manager software. Infrared radiation was reflected by markers on the buoys and captured by high speed cam-eras. The exact position was calculated using triangulation. Tension in the chain was recorded using submersible load cells with regular strain gauges. Sensors were placed in ev-ery chain close to the buoy and the anchor. These sensors measured the force deviation relative to the force present in quiescent water. The total force is then equal to the summa-tion of the pretension and the measured forces due to wave action. The pretension was calculated with a catenary equa-tion according to Jonkman and Buhl [9]. The results were 0.9674 N and 1.588 N at the location of the anchor and buoy sensors, respectively. The simulated pretension was also as-sessed by performing a simulation without waves. The val-ues amounted to0.9225 N at the anchor sensor location and 1.4125 N at the location of the buoy sensor. The pretension force makes up the largest part of the forces. The pretension predicted by the numerical model is used to calculate the total tension. This way, the forces originating from wave action can be compared more easily. The catman software by HBM was used for the post-processing of the results. A sketch of the used model is shown in figure 1. The main dimensions are given as well.

The set-up was modelled as closely as possible in the MoorDyn-UGent software. There were some discrepancies between both. For example, the way of connection between the components was not modelled. The pitch motion ob-served during testing was not considered in the numerical calculations either. The values for the drag and added mass coefficients follow the guidelines of OrcaFlex for circular cylinders.

Figure 1: CAD drawing of the model (lengths given in mm).

IV. Results

The waves that were used for the validation have a wave pe-riod between 1 and2 s and a wave height ranging from 10 to 30 cm. Other waves were not generated or showed a highly irregular character. For a wave period of1 s, all wave heights were generated. For a wave period of1.5 s, wave heights up to25 cm were generated. For a wave period of 2 s, wave heights up to20 cm were generated. The comparison of the results is made per incoming wave angle. The maximum val-ues of the forces are used for the comparison. They are given as a function of the wave height for different wave periods. The position measurements are more difficult to interpret. Imperfections in the set-up and the significant pitch motion which isn’t modelled cause large differences between mea-sured and simulated results. General trends can be observed, but the absolute values are seldom comparable. The proper-ties used for the comparison are the ranges of coordinates in the three directions. Chain 1, buoy 1 and anchor 1 denote the respective items closest to the wave maker.

A. 0 degrees

Figure 2 shows the measured and modelled forces for a wave period of1.5 s. These results are exemplary for the tests with 0°. The higher forces are over-predicted by the model. A good agreement is found for the lower values. This over-estimation might be the result of a faulty force sensor cal-ibration or dynamic effects not included in the numerical model. The higher values in chain 1 could also be caused by a higher pretension. During the experiments, the catenary shape changes significantly and leads to a higher pretension. The experimental results however use a constant value for the pretension. This does not explain the over-prediction for forces in the second chain, observed in the tests with a period of1 s.

The position ranges during the same tests are shown in figure 3. The motion in the x-direction of the first buoy is well predicted, with a maximum relative error of 10 %. The second buoy moves significantly more in the x-direction than the predicted motion. This is most likely the conse-quence of a bad calibration of the positioning equipment be-fore these tests. Due to the symmetrical nature of the set-up, no y-motion is predicted. Imperfections in the physical model lead to a small motion along the y-axis. As was the case for most experiments, the measured heave motion ex-ceeded the simulated motion. A flexibility in the connection between the buoys and other components might be the rea-son for this behaviour.

(8)

(a) Anchor 1 (b) Anchor2

(c) Buoy 1 (d) Buoy 2

Figure 2: Measured and simulated forces for waves with a wave period of1.5 s and incoming wave angle of 0°.

(a) Buoy 1, x (b) Buoy 2, x

(c) Buoy 1, y (d) Buoy 2, y

(e) Buoy 1, z (f) Buoy 2, z

Figure 3: Measured and simulated position ranges for waves with a wave period of1.5 s and incoming wave angle of 0°. B. 30 degrees

During these tests, the sensor at the location of buoy 1 broke. The simulated values are given for completeness.

The maximum measured and simulated forces during the tests with a period of1.5 s are depicted in figure 4. They summarize the trends observed for this wave angle. The low forces, observed here in the second chain and during tests with wave period2 s, are predicted well. An over-prediction of the model is found for the higher forces.

The position ranges of the same tests are shown in figure 5. The motion in the x-direction is most often over-predicted by MoorDyn-UGent. The reverse is true for the heave

mo-(a) Anchor 1 (b) Anchor2

(c) Buoy 1 (d) Buoy 2

Figure 4: Measured and simulated forces for waves with a wave period of1.5 s and incoming wave angle of 30°.

tion of the buoys. The range of the y-positions is predicted relatively well for the tests with an incoming wave angle of 30°.

(a) Buoy 1, x (b) Buoy 2, x

(c) Buoy 1, y (d) Buoy 2, y

(e) Buoy 1, z (f) Buoy 2, z

Figure 5: Measured and simulated position ranges for waves with a wave period of1.5 s and incoming wave angle of 30°.

C. 60 degrees

The measured and simulated forces match closely for the ex-periments featuring an incoming wave angle of 60°. Less accurate results for the tests with a wave height of1.5 are probably the effect of a fault in the set-up.

The same trends as above hold for the motions in the x-and z- directions. No clear trend was observed regarding the recorded y-values.

(9)

Figure 6 shows the time series of the forces during the test with a wave height of15 cm and a wave period of 2 s. The agreement is good, except for the slightly different build-up stage.

(a) Anchor 1 (b) Anchor2

(c) Buoy 1 (d) Buoy 2

Figure 6: Time series of the forces with a wave period of2 s, a wave height of15 cm and incoming wave angle of 60°.

D. 90 degrees

The forces present during the tests with a wave period of 1 s illustrate the behaviour for waves perpendicular to the set-up (see figure 7). A good agreement is found between simulated and measured values. Only during the tests with a period of2 s an overestimation was observed.

(a) Anchor 1 (b) Anchor2

(c) Buoy 1 (d) Buoy 2

Figure 7: Measured and simulated forces for waves with a wave period of1 s and incoming wave angle of 90°.

The trends for the motions during the tests with waves perpendicular to the longline can be observed in figure 8. The motions in the x-direction are overestimated, while the heave motion of the buoys is underestimated. The predicted values for the motions in the y-direction are small; the mea-sured values are slightly higher.

(a) Buoy 1, x (b) Buoy 2, x

(c) Buoy 1, y (d) Buoy 2, y

(e) Buoy 1, z (f) Buoy 2, z

Figure 8: Measured and simulated position ranges for waves with a wave period of1.5 s and incoming wave angle of 90°. V. Conclusion and future work

The measured and simulated forces show a better agreement the less the model is aligned with the waves. The simulations for0° often overestimate the real tension in the chain. A pos-sible, yet unlikely cause could be a faulty sensor calibration during these tests. Another possibility lies in the assumption of constant pretension values. During testing, the catenary shape of the first chain changes significantly and the pre-tension load with it. This is not included in the measured results. The neglect of some dynamic phenomena can also be a reason, for example the flow reduction at the mussel droppers. The numerical model inevitably shows some dif-ferences with the physical set-up. The way of connecting the components is not implemented in the software, yet leads to some flexibility in the longline model. Further investigation of the time series showed that the higher pretension explains only part of the higher values.

Few conclusions can be drawn from the position mea-surements. The data show a large spread and often signifi-cant differences between measured and simulated outcomes. Two trends can be observed. The predicted motion in the x-direction exceeds the measured values in almost every case. The reverse holds true for the heave motion. Nevertheless, an accurate prediction of the positions is primordial to as-sess the forces. A higher prediction of the motions in the x-direction also leads to a higher prediction of force oscilla-tions.

Additional validation projects are needed to assess the MoorDyn-UGent accuracy. Higher regular waves as well as irregular waves should be simulated, combined with cur-rent profiles. Accurate modelling of real mussel dropper lines is needed when simulating mussel farms. The

(10)

mus-sel lines constitute the principal drag element and have an influence on the total load of the longline. Flow reduction should also be included. At the time of writing, a version of MoorDyn-UGent with 6DOF motion for the buoys is be-ing developed. Better agreement between the predicted and measured positions would significantly increase the accu-racy for the force predictions as well. After sufficient vali-dation, the MoorDyn-UGent software can be used to design the optimal configuration of a mussel farm in harsh weather conditions.

Bibliography

[1] D. L. Aksnes, P. Holm, M. Bavinck, F. Biermann, R. Donovaro, P. Harvey, S. Hynes, J. Ingram, M. Kaiser, S. Kaushik et al., Food from the Oceans-How can more

food and biomass be obtained from the oceans in a way that does not deprive future generations of their benefits?

Science Advice for Policy by European Academies (SA-PEA), 2017.

[2] Ó. Ögmundarson, J. Holmyard, G. Þórðarson, F. Sig-urðsson, and H. Gunnlaugsdóttir, “Offshore aquacul-ture farming–report from the initial feasibility study and market requirements for the innovations from the project,” Icelandic Food and Biotech, Reykjavík, 2011. [3] R. Langan and F. Horton, “Design, operation and

eco-nomics of submerged longline mussel culture in the open ocean,” Bulletin of the Aquaculture Association of

Canada, vol. 103, no. 3, pp. 11–20, 2003.

[4] B. Buck, “Experimental trials on the feasibility of off-shore seed production of the mussel Mytilus edulis in the German Bight: installation, technical requirements and environmental conditions,” Helgol Mar Res, vol. 61, pp. 87–101, 2007.

[5] D. Morse and M. A. Rice, “Mussel aquaculture in the northeast,” Northeastern Regional Aquaculture Center

Publication, vol. 211, no. 10, 2010.

[6] B. H. Buck, A. Berg-Pollack, J. Assheuer, O. Zielinski, and D. Kassen, “Technical realization of extensive aqua-culture constructions in offshore wind farms: consid-eration of the mechanical loads,” in 25th International

Conference on Offshore Mechanics and Arctic Engineering.

American Society of Mechanical Engineers Digital Col-lection, 2006, pp. 691–697.

[7] S. Lagerveld, C. Rockmann, M. Scholl, H. Bartel-ings, S. van den Burg, R. Jak, H. Jansen, J. Klijnstra, M. Leopold, M. Poelman et al., “Combining offshore wind energy and large-scale mussel farming: back-ground & technical, ecological and economic consider-ations,” IMARES, Tech. Rep., 2014.

[8] M. Hall and A. Goupee, “Validation of a lumped-mass mooring line model with DeepCwind semisubmersible model test data,” Ocean Engineering, vol. 104, pp. 590– 603, 2015.

[9] J. Jonkman and M. Buhl, “Development and verification of a fully coupled simulator for offshore wind turbines,”

in 45th AIAA Aerospace Sciences Meeting and Exhibit, 2007, p. 212.

(11)

Contents

Acknowledgements ii

Abstract iii

List of Figures xi

List of Tables xiv

1 Introduction 1 1.1 Aquaculture . . . 1 1.2 Co-use of area . . . 3 1.3 Belgium . . . 6 2 Theoretical background 8 2.1 MoorDyn . . . 8 2.1.1 Original MoorDyn . . . 8 2.1.2 MoorDyn-UGent . . . 15 2.2 Experimental modelling . . . 18 3 Methodology 21 3.1 Model tests . . . 21 3.1.1 MC-Lab . . . 21 3.1.2 Model . . . 26 3.2 MoorDyn . . . 32 ix

(12)

4 Results 34 4.1 0 degrees . . . 35 4.1.1 Results . . . 35 4.1.2 Discussion . . . 38 4.2 30 degrees . . . 41 4.2.1 Results . . . 41 4.2.2 Discussion . . . 48 4.3 60 degrees . . . 50 4.3.1 Results . . . 50 4.3.2 Discussion . . . 55 4.4 90 degrees . . . 55 4.4.1 Results . . . 55 4.4.2 Discussion . . . 58

5 Conclusion and future work 63 5.1 Conclusion . . . 63

5.2 Future work . . . 64

A Force sensor data sheets 66

B MoorDyn-UGent input file 72

(13)

List of Figures

1.1 Surface longline. (Ögmundarson et al., 2011) . . . 2

1.2 Semi-submerged longline. (Ögmundarson et al., 2011) . . . 2

1.3 Submerged longline. (Ögmundarson et al., 2011) . . . 3

1.4 Location of wind farms in the Belgian part of the North Sea. (IMC, 2012). . . 5

1.5 Force line deployed in Belwind area. (ILVO and Ghent University, 2017) . . . 7

2.1 Mooring line discretisation and indexing. (Hall, 2017b) . . . 9

2.2 Forces on a line. (Hall and Goupee, 2015) . . . 10

2.3 MoorDyn input file. (Hall, 2017b) . . . 12

2.4 Wave input for MoorDyn-UGent. . . 18

2.5 Post-processing options in MoorDyn-Ugent. . . 19

3.1 Test section of the wave basin in the MC-Lab. . . 22

3.2 Position of the Oqus cameras used for motion capture. . . 24

3.3 Edulis test line in Belwind. (Pribadi et al., 2019) . . . 26

3.4 CAD drawing of the model (lengths given in mm). . . 27

3.5 Deployed model ready for testing. . . 28

3.6 Anchor and force sensor. . . 29

3.7 Tensile test set-up. . . 29

3.8 Stress-strain curve of the backbone material. . . 30

3.9 Connection of a dropper to the backbone. . . 30

3.10 Buoy used in the experiments. . . 31 xi

(14)

3.11 MoorDyn Line Types section. . . 32 3.12 MoorDyn Connection Properties section. . . 33 3.13 MoorDyn Solver Options section. . . 33 4.1 Measured and simulated forces for waves with a wave period of1 s and

incom-ing wave angle of0°. . . 36 4.2 Measured and simulated position ranges for waves with a wave period of1 s

and incoming wave angle of0°. . . 37 4.3 Measured and simulated forces for waves with a wave period of1.5 s and

in-coming wave angle of0°. . . 38 4.4 Time series of the forces with a wave period of1.5 s, a wave height of 15 cm and

incoming wave angle of0°. . . 39 4.5 Measured and simulated position ranges for waves with a wave period of1.5 s

and incoming wave angle of0°. . . 40 4.6 Measured and simulated forces for waves with a wave period of2 s and

incom-ing wave angle of0°. . . 41 4.7 Measured and simulated position ranges for waves with a wave period of2 s

and incoming wave angle of0°. . . 42 4.8 Measured and simulated forces for waves with a wave period of1 s and

incom-ing wave angle of30°. . . 43 4.9 Measured and simulated position ranges for waves with a wave period of1 s

and incoming wave angle of30°. . . 44 4.10 Measured and simulated forces for waves with a wave period of1.5 s and

in-coming wave angle of30°. . . 45 4.11 Measured and simulated position ranges for waves with a wave period of1.5 s

and incoming wave angle of30°. . . 46 4.12 Time series of the buoy positions for waves with a wave period of1.5 s, a wave

height of15 cm and incoming wave angle of 30°. . . 47 4.13 Measured and simulated forces for waves with a wave period of2 s and

incom-ing wave angle of30°. . . 48 4.14 Measured and simulated position ranges for waves with a wave period of2 s

(15)

4.15 Measured and simulated forces for waves with a wave period of1 s and incom-ing wave angle of60°. . . 50 4.16 Measured and simulated position ranges for waves with a wave period of1 s

and incoming wave angle of60°. . . 51 4.17 Measured and simulated forces for waves with a wave period of1.5 s and

in-coming wave angle of60°. . . 52 4.18 Measured and simulated position ranges for waves with a wave period of1.5 s

and incoming wave angle of60°. . . 53 4.19 Time series of the forces with a wave period of2 s, a wave height of 15 cm and

incoming wave angle of60°. . . 54 4.20 Measured and simulated forces for waves with a wave period of1 s and

incom-ing wave angle of90°. . . 56 4.21 Measured and simulated position ranges for waves with a wave period of1 s

and incoming wave angle of90°. . . 57 4.22 Measured and simulated forces for waves with a wave period of1.5 s and

in-coming wave angle of90°. . . 58 4.23 Measured and simulated position ranges for waves with a wave period of1.5 s

and incoming wave angle of90°. . . 59 4.24 Time series of the buoy positions for waves with a wave period of1.5 s, a wave

height of15 cm and incoming wave angle of 90°. . . 60 4.25 Measured and simulated forces for waves with a wave period of2 s and

incom-ing wave angle of90°. . . 61 4.26 Measured and simulated position ranges for waves with a wave period of2 s

(16)

List of Tables

2.1 Scaling factors according to Froude scaling. . . 20 3.1 Line properties of the Belwind test line. (Pribadi et al., 2019) . . . 26 3.2 Buoy properties of the Belwind test line. (Pribadi et al., 2019) . . . 26 4.1 Measured and simulated forces for waves with a wave period of2 s and

incom-ing wave angle of60°. . . 54 4.2 Measured and simulated position ranges in cm for waves with a wave period of

2 s and incoming wave angle of 60°. . . 55

(17)

Chapter 1

Introduction

1.1 Aquaculture

Aquaculture is the fastest growing food production sector in the world with an average annual growth rate of5.8 % between 2000 and 2016 (Food and Agriculture Organization, 2018). A pub-lication of the European Union mentions mariculture, or aquaculture of marine species, as the sector with the greatest potential to answer the growing demand for food (Aksnes et al., 2017). As part of the Blue Growth strategy, the European Commission promotes the sustainable devel-opment of aquaculture issuing several directives. The term aquaculture comprises the farming of fish, crustaceans, aquatic plants, algae and molluscs.

Mussel exploitation in Europe has almost exclusively been practised in protected near-shore waters. These areas are limited and attract many stakeholders, which limits the development of coastal aquaculture. Aquaculture farther offshore is believed to have enormous potential (Walter et al., 2002). Another advantage of cultivating species farther from the coast is the open ocean’s better water quality. Offshore farms also tend to have a better growth rate and shorter production cycles (Ögmundarson et al., 2011). "Offshore" in the aquaculture sector is defined as being in an area fully exposed to all kinds of environmental conditions.

The main prerequisite for the success of offshore systems is the ability to survive the harsh conditions. Of the techniques commonly practised, several studies illustrate longlines as the most promising technique (Danioux et al., 2000; Langan and Horton, 2003; Buck, 2007; Langan, 2013). Compared to other techniques like rafts or on-bottom types they can be easily adapted to deeper waters and more exposed sites (Morse and Rice, 2010).

In longline culture, substrates with on-growing species are suspended from a horizontal rope 1

(18)

Chapter 1. Introduction 2 called the backbone. The dropper lines can be discontinuous vertical ropes or one continuous rope that forms a V-pattern. Buoys are connected to the line for buoyancy and the longline is moored with anchors and chains. There are three types of longlines: floating, semi-submerged and submerged lines (Ögmundarson et al., 2011).

Floating longlines are placed directly at the water surface and follow the motion of the waves.

This makes them unsuitable for heavy weather operations. The considerable motions cause big forces in the lines and an increased chance of mussels dropping off (Ögmundarson et al., 2011). The effect of the waves becomes less the more the longline is submerged. Figure 1.1 provides a sketch of a surface longline.

Figure 1.1: Surface longline. (Ögmundarson et al., 2011)

In a semi-submerged longline the longline is situated several metres below the surface. This reduces the motion and the loads on the lines. The floats are placed directly at the line and at the surface. A semi-submerged line is shown in figure 1.2.

Figure 1.2: Semi-submerged longline. (Ögmundarson et al., 2011)

The submerged longline can only be discerned at the surface by a tracking float. The buoyancy floats are attached to the longline and are submerged. The advantage of this system is that it is placed entirely below the damaging effects of the weather. It is however more difficult to carry out regular inspections and to keep the droppers off the bottom (Ögmundarson et al., 2011). An

(19)

1.2. Co-use of area 3 example of a submerged longline is shown in figure 1.3.

Figure 1.3: Submerged longline. (Ögmundarson et al., 2011)

Danioux et al. describe the situation of offshore mussel cultivation in the Mediterranean Sea in the year 2000 (Danioux et al., 2000). France and Italy were the largest producers, while other countries were carrying out tests. The technical aspects of longlines are discussed and an economic analysis is made. Between 1988 and 1991 IFREMER (Institut français de recherche pour l’exploitation de la mer; French Research Institute for Exploitation of the Sea) developed the semi-submerged line as a compromise between the floating lines and submerged lines. Several projects have assessed the economic, biological and technical expediency of offshore mussel aquaculture. In the German Bight for example, the feasibility study led to positive re-sults (Buck, 2007). According to the biological study, this offshore test location was suitable as an offshore seed production site as well as a grow-out site. Two submerged longlines were de-ployed, of which the polypropylene one showed the greatest potential. A 2010 economic study by Buck et al. showed that offshore mussel production for consumption is profitable (Buck et al., 2010). Another project deployed two submerged longlines 10 km off the Portsmouth coast and tested the operational and commercial feasibility of blue mussel (Mytilus edulis) cultivation (Langan and Horton, 2003).

1.2 Co-use of area

A lot of parties are involved in the exploitation of the Belgian part of the North Sea. Activi-ties taking place in the area include scientific research, shipping, commercial fishing, marine aquaculture and tourism. Moreover, some zones are reserved for the generation of renewable energy. The Marine Spatial Plan (MSP) settles the division of the area between the parties in-volved (Belgian Government, 2020). The limited area sometimes leads to conflicts of interest.

(20)

Chapter 1. Introduction 4 One of the possibilities to cope with this is the co-use of the area. The 2014 MSP preserved the possibility of sustainable mariculture in the wind turbine concession zones of Belwind I and C-Power. The location of these zones is shown in figure 1.4. In the 2020 MSP more zones are indicated as possible locations for mariculture (Bossier et al., 2018). For a complete overview on aquaculture in Belgium and the corresponding legislation, the reader is referred to the sum-mary made by the Flemish Institute for the Sea (Vlaams Instituut voor de Zee, VLIZ) (Bossier et al., 2018).

Around the world projects have started to check the feasibility of merging offshore wind farms and mussel aquaculture. Advantages of implementing mussel cultivation into wind farm terri-tories include space and cost savings, attachment possibilities for the mariculture facilities and sharing of resources (Michler-Cieluch et al., 2009b). The regular servicing of the turbines can for example be combined with the harvesting and maintenance of the mussel longlines. Overall, this combination would allow aquaculture to be executed in a more sustainable and efficient way. A 2006 study by Buck et al. analyses the technical possibility of using the offshore founda-tion structures as a fixafounda-tion device for longlines (Buck et al., 2006). It assumes the economically sound installation of longlines is not possible without the foundations of the wind turbines as anchor points. Experimental data on longline forces induced by environmental actions was retrieved from a test line in the German North Sea. Several possibilities for the organisational structure of the cooperation are discussed by Krause et al. (Krause et al., 2011). The Blauw-druk project in the Netherlands examined the technical, ecological and economic feasibility of combining large-scale mussel farming and offshore wind energy production in the North Sea. Its focus was on the1000 MW wind farm concessions on the Dutch Continental Shelf. A 2014 report by Lagerveld et al. summarizes the results (Lagerveld et al., 2014). A thorough overview of all opportunities and challenges is given by Michler-Cieluch (Michler-Cieluch et al., 2009a). A list of all feasibility studies performed on offshore aquaculture and co-use of area was made by Buck et al. (Buck and Krause, 2013). However, the likelihood of a successful cooperation not only depends on feasibility studies, but on the social environment and regulatory frameworks as well (Buck et al., 2004). These frameworks can stimulate or counteract the realisation of multi-use concepts (Jansen et al., 2016).

(21)

1.2. Co-use of area 5

(22)

Chapter 1. Introduction 6

1.3 Belgium

Several companies and scientific institutes are investigating the development of sustainable aquaculture in Flanders and the Belgian part of the North Sea (BNS). The feasibility of co-use of area was first assessed in the 2015 MARIPAS-project. In 2014-2015 the AquaValue-project developed a roadmap leading to integrated aquaculture at sea. Four potential activities were identified, one of which was mussel cultivation in wind farms. The feasibility of off-bottom mussel cultivation in the Belgian wind farms was assessed by the Norwegian research institute SINTEF. The ’Noordzee Aquacultuur’ research project (North Sea Aquaculture) was initiated next (ILVO and Ghent University, 2017). Reseach institutes and industrial partners are work-ing together in this project: ILVO, Ghent University, OD Natuur, C-Power, Belwind, DEME, Sioen Industries, Colruyt Group, Brevisco and Lobster Fish. It consists of two independent projects: the Value@Sea project near Nieuwpoort and the Edulis project in the Belgian wind farms C-Power and Belwind. The Value@Sea project investigates the technical, ecological and economical feasibility of the integrated culture of the European flat oyster, the great scallop and sugar kelp.

The Edulis project assesses the ecological and technical feasibility of combining offshore wind parks with blue mussel aquaculture using the longline technique. An economical analysis is made as well to see if this can be done in a profitable way. The ecological feasibility is assessed by the Laboratory of Aquaculture & ARC of the Faculty of Bioscience Engineering. The tech-nical aspect comprises the design of the lines and the modelling of the forces in and positions of the system. This is done by the Maritime Technology Division (MTD) of the Department of Civil Engineering, sub-unit of the Faculty of Engineering and Architecture. For this purpose the MTD developed the numerical program MoorDyn-UGent. This software is used to model the system and its reaction to environmental loads. If the program is sufficiently validated, it can be used to optimize a configuration that can withstand the environmental loads of the North Sea.

Two test lines have already been deployed within the Edulis project: a "Bio line" in the C-Power concession zone and a "Force line" in the Belwind area. The Bio line was used to check the spatfall, the growth of the mussels and the optimal type of rope. The Force line was a longline with fully grown mussels and was accommodated with force sensors on the chains. The forces and positions of the system were measured and a first validation of the MoorDyn-UGent program was done. So far, the results of this validation have only been disclosed to the Edulis project partners. A sketch of the Force line is demonstrated in figure 1.5.

(23)

1.3. Belgium 7

Figure 1.5: Force line deployed in Belwind area. (ILVO and Ghent University, 2017)

scale model of a longline is built and subject to a range of incoming waves. Tension forces in the chains and buoy positions are measured. These tests are performed at NTNU in Trond-heim, Norway. The results of these measurements are then compared with the outcome of the MoorDyn-UGent simulations.

(24)

Chapter 2

Theoretical background

2.1 MoorDyn

The set-up is simulated in the MoorDyn-UGent software. MoorDyn is an open-source lumped-mass mooring line model developed by Matthew Hall. It can be used to calculate the dynamics of mooring systems. The model was validated for the first time in 2015 by Hall and Goupee (Hall and Goupee, 2015). Test data from a scaled floating offshore wind turbine was used for the validation. Another validation was done by Vissio et al. with tests performed on a 1:20 scaled wave energy converter (Vissio et al., 2015). The results of both validation projects suggested that the MoorDyn model was computationally efficient and suitable for predicting the mooring line loads. The efficiency is apparent in the good accuracy obtained with a coarse discretisation. To increase the versatility, means to include seabed friction and the possibility to model multiple floating bodies were added in 2017 (Hall, 2017a). A practical user’s guide was published in 2017 (Hall, 2017b). There are a C++ and a FORTRAN version of MoorDyn. The difference between both lies in the implementation. The Maritime Technology Division of Ghent University (MTD) modified the FORTRAN version to include environmental inputs (Pribadi and Donatini, 2018b). In this section the main principles of the software will be explained. First, the original MoorDyn is looked at. The adaptations made by the MTD are discussed next.

2.1.1 Original MoorDyn

The basic mathematical principles of the MoorDyn code are given in this section. The reader is referred to the MoorDyn User’s Guide by Matthew Hall (Hall, 2017b) and the validation publication by Hall and Goupee (Hall and Goupee, 2015) for more information.

(25)

2.1. MoorDyn 9

Model structure

MoorDyn uses a lumped-mass approach. To this end the lines are divided into N evenly-sized segments and N+1 nodes. The mass of the segments is transferred to the adjacent nodes. The nodes get an index between 0 and N; the cable segment between nodes 0 and 1 is given an index of 1/2. Every segment of a line has the same properties: unstretched length, diameter, density and Young’s modulus. A right-handed inertial reference frame is used and the vectorr denotes

the position of every node. These conventions are shown in figure 2.1.

Figure 2.1: Mooring line discretisation and indexing. (Hall, 2017b)

The model only considers axial stiffness and thus neglects bending and torsional stiffness. These two are of minimum importance in chain-based systems, so their neglect is permitted (Vissio et al., 2015). MoorDyn includes damping forces, weight, buoyancy forces, contact forces from the seabed and hydrodynamic forces according to Morison’s equation (Hall and Goupee, 2015). These forces are depicted in figure 2.2. The model does not account for compression forces so the tension force T is only applied if it’s positive. Contrary to most other programs, MoorDyn calculates the loads at the nodes and not at the segment midpoints. The dotted line in the figure shows the tangent direction to the line at the node point. This tangent, necessary for calculating the hydrodynamic loads, is approximated as the average of the tangent directions to the adjacent line segments. In the original MoorDyn code, the effects of wave kinematics are neglected. The omission of these effects is justified since the results show a maximum deviation of 3% when wave loads are included. The MTD will include the wave kinematics in their version. For more information regarding the force calculations in MoorDyn, the reader is referred to the validation publication by Hall and Goupee (Hall and Goupee, 2015).

(26)

Chapter 2. Theoretical background 10

Figure 2.2: Forces on a line. (Hall and Goupee, 2015)

The end of every line is defined by a connection object. There are three types of connection nodes:

• Fixed nodes are fixed in space. Anchor points are modelled as fixed nodes.

• Vessel nodes undergo a predefined motion. This motion is mostly given by a different simulation program and represents the motion of fairlead connections.

• Connect nodes move according to the forces acting on them. These forces can be forces exerted by attached mooring lines or external forces. Connect nodes are used to join two or more lines.

Model operation

The operation of the model can be divided into multiple steps. In the initialisation step, the input file describing the system is read and the initial equilibrium state is determined. Two stages can be discerned in the determination process.

• First, the location of all the points along the lines is determined using a quasi-static model. The position of the end points of the lines is known from the input file. The intermedi-ate points are locintermedi-ated using a cintermedi-atenary equation, only taking into account the mass and

(27)

2.1. MoorDyn 11 buoyancy of lines. The tension force at both ends of a line is also calculated in this stage (Jonkman and Buhl, 2007).

• In the second stage the correct position of all points is calculated. The mooring system settles to an equilibrium using dynamic relaxation with boosted drag coefficients. The first guess for the position of the connect nodes in the input file is also corrected. In this stage the model also takes into account the weight of connection objects.

Following the initialisation step are a number of coupling time steps. During each step, Mo-orDyn calculates the forces on and accelerations of all the nodes. The equation of motion for every node is given by equation 2.1, where𝑟 denotes the position and ̈𝑟 the acceleration (Pribadi and Donatini, 2018a):

(𝑚 + 𝑎) ̈𝑟 = 𝐹(𝑟, ̇𝑟, 𝑡) (2.1)

With F:

𝐹 = 𝐷𝑝+ 𝐷𝑞+ 𝑇 + 𝐶 + 𝑊 + 𝐵 (2.2)

The internal forces are the axial stiffness (𝑇), line damping force(𝐶) and net weight (𝑊 ). The vertical bottom contact (𝐵) is modelled as a spring-damper system. Hydrodynamic forces are split into transverse (𝐷𝑝) and tangential (𝐷𝑞) drag forces. As mentioned before, the wave

kine-matics are not modelled.

The second order differential equation (2.1) at every node is then reduced to a first order problem by introducing state vectors. The first order differential equations are solved with a Runge-Kutta second order integration scheme. A detailed formulation of the method can be found in the Theoretical Manual by Pribadi and Donatini (Pribadi and Donatini, 2018a). Depending on the settings defined by the user, certain quantities are printed to the output.

The termination step finishes the calculations and the output files are closed.

Mooring system

The entire mooring system is described in one input file: a DATA-file named MoorDyn in the FORTRAN version. The example input file from the MoorDyn User’s Guide (Hall, 2017b) will be used here to denote the different parts. The structure will later be applied to the mussel longline in the Methodology chapter. The example input file is shown in figure 2.3.

(28)

Chapter 2. Theoretical background 12

Figure 2.3: MoorDyn input file. (Hall, 2017b)

TheLine Types section contains the properties of all the lines that are used in the system. The

system in the example uses only one line type "main". The properties needed to stipulate the line are:

• Name: identifier for the line type

• Diam (m): volume-equivalent diameter of the line; MoorDyn treats all lines as cylinders • MassDen (kg/m): mass per unit of length

(29)

2.1. MoorDyn 13 • BA/−𝜁 (N-s/-): line internal damping; if a negative value is entered the value is read as

the desired damping ratio as a fraction of the critical damping • Can (-): transverse added mass coefficient

• Cat (-): tangential added mass coefficient • Cdn (-): transverse drag coefficient • Cdt (-): tangential drag coefficient

The Connection Properties section declares all the connection nodes. Every node has the

following characteristics:

• Node: every node is given a number • Type: Fixed, Vessel or Connect • X,Y,Z (m): node coordinates

• M (kg): node mass if it denotes a clump weight • V (m3): node displacement in the case of floats

• FX, FY, FZ (kN): external force components acting on the node • CdA (m2): product of drag coefficient and projected area

• Ca (-): added mass coefficient

TheLine Properties section defines all the lines to be simulated. The line type is specified, as

are the length of the line and the number of segments it consists of. The end points of all lines are also chosen. The last column denotes the results that get a separate output file. The output file for a given line contains the chosen output property at each node/segment. The possibilities for the output properties are:

• p: node position • v: node velocity • U: wave velocity

(30)

Chapter 2. Theoretical background 14 • t: tension force

• c: internal damping force • s: strain

• d: rate of strain

The Solver Options section lists a number of settings specified by the user. Default settings

are used if a line is missing. The possible settings are: • g (m/s2): gravitational constant

• rhoW (kg/m3): water density

• dtM (s): mooring integration time step • kBot (Pa/m): bottom stiffness

• cBot (Pa-s/m): bottom damping • WtrDpth (m): water depth

• dtIC (s): time interval to assess convergence during Initial Conditions generation • TmaxIC (s): max time for IC generation

• CdScaleIC (-): scale factor for drag coefficients during dynamic relaxation • threshIC (-): threshold for IC convergence

The Outputs section lists the general outputs that are printed to the global output file. The

possible output quantities are: • pX,pY,pZ (m): coordinates • vX,vY,vZ (m/s): velocities • aX,aY,aZ (𝑚/𝑠2): accelerations

• T/Ten (N): tension • fX,fY,fZ (N): net force

These quantities can be produced at a connection object by using the prefix ’Con#’, # being the connect number. They can also be produced at a node along a line with the prefix ’L#N@’, with # the line number and @ the number of the node along that line.

(31)

2.1. MoorDyn 15

2.1.2 MoorDyn-UGent

In this section the mathematical principles of the MoorDyn-UGent software are discussed. These principles and more can be found in the technical manual by Pribadi and Donatini (Prib-adi and Donatini, 2018a). It mostly concerns adaptations made to the original MoorDyn code. The main difference is the inclusion of environmental effects; a prescribed motion of the fair-leads is no longer necessary.

As mentioned above, the MoorDyn-UGent software is being validated using full-scale measure-ments on the Edulis test line in the Belwind concession zone. Results of a numerical verification have already been published (Pribadi et al., 2019). A simple set-up of an anchor connected to a buoy through a chain was modelled in both MoorDyn-UGent and OrcaFlex, a commercially available lumped-mass mooring line model. The results of both simulations were compared. There was a good agreement between the predicted positions of the buoy in all test cases. The magnitude of snap loads however showed some discrepancy. The real magnitude of the snap loads should be determined experimentally.

Environmental loads

The hydrodynamic force calculation of MoorDyn-UGent accounts for the fluid velocity and acceleration. This leads to the addition of the Froude-Krylov force and the added mass force from the fluid acceleration. The relative velocity between body and fluid is used to calculate the drag forces.

By providing a current speed𝑣0at a reference depth𝑧0and an exponent𝛼, a current profile is

calculated:

𝑣 = 𝑣0(𝑑 + 𝑧𝑑 + 𝑧 0)

𝛼

(2.3) The direction of the current is also defined by the user in the input file. No current was applied in the tests executed for this project so no current is modelled in MoorDyn-UGent.

A regular wave is modelled by giving its amplitude, direction, period and phase. Linear Airy wave theory is used to calculate the wave elevation and kinematics. Since the Airy theory is only valid below the calm water surface Wheeler’s kinematic stretching approach is used. The exact formulas can be found in the Theoretical Manual for MoorDyn-UGent (Pribadi and Donatini, 2018a).

(32)

Chapter 2. Theoretical background 16 Irregular waves are composed of multiple regular waves with specific amplitude, frequency and direction. The user also assigns phases to all the components to get a final sea state. With the assumption of linear behaviour the superposition principle can be applied and the total loads are given by the summation of the contribution of the components.

A final addition involving the environmental loads is the possibility of a build up stage. When a ramp up time 𝑑0 is given the environmental loads undergo a gradual increase following a

fifth-order Smoothstep function.

Line seabed friction

The line seabed friction can also be modelled with MoorDyn-UGent. In the experimental set-up however, there is no contact between any lines and the bottom. Therefore the friction force is neglected and no detailed description will be given. For more information the reader is referred to the aforementioned journal paper by Pribadi et al. (Pribadi et al., 2019).

Buoy and clump weight modelling

In the original MoorDyn software, it is possible to add mass and volume to a connection. The exact shape however remains unknown and the hydrodynamic behaviour of buoys cannot be calculated. In the version modified by the MTD the width and the height of the buoy can also be chosen. Buoys are assumed cylindrical and always in an upright position. The bottom of the buoy is at the connection node. The node’s position is used to get the free surface and to calculate the wave kinematics. Larger diameters will thus lead to a less accurate result.

Clump weights are modelled in the same way as buoys. Static and kinetic friction coefficients are also assigned to the node. This is of less importance in the present study since the anchors are fixed and no other weights are present.

Utilisation

The utilisation of the MoorDyn-UGent software is much the same as the original MoorDyn. Only small adaptations were made to the input file. In this section a short description of these changes will be given. More information about how to use the MoorDyn-UGent software can be found in the MoorDyn-UGent User’s Manual (Pribadi and Donatini, 2018b).

(33)

2.1. MoorDyn 17 Compared to the original MoorDyn code, there are four additional properties in theLine type

section of the input file. These properties are the static and kinetic friction coefficients in both longitudinal and transverse direction [-].

In theConnection properties section there are some new properties while others request a

different input form. In the case of a buoy or clump weight the added mass and drag coefficients in the normal and tangential direction [-] are now all given separately. As mentioned before the height and width [metre] of buoys and clump weights are now also needed in the input file. For a clump weight the same four friction coefficients apply as for the line types.

The sole difference in theLine properties section is the meaning of ’U’ in the output column.

If ’U’ is chosen the added mass and Froude-Krylov forces for every segment of that line will be printed to an output file.

There are a number of additions in theSolver options section:

• dt0 (s): time step in which the output files will be produced • Tend (s): End of the simulation time

• CurrV (m/s): current speed at the reference depth • CurrZ (m): reference depth

• CurrDir (deg): current direction

• CurrAlpha (-): exponent for the power law profile

• WaveSwitch (-): 0 if no waves are present, 1 if a wave is to be included • d0 (s): duration of the build up stage

In MoorDyn-UGent a separate text file is required for the wave input. Figure 2.4 illustrates the input of the wave properties. The first line contains the number of wave periods and wave directions. These are both equal to 1 for a regular wave. The next number denotes the wave period in seconds (multiple rows are used in case of multiple periods). The wave direction(s) are given in the same way. The last numbers define the amplitude (inm) and phase of the wave. Special attention should be given to the convention used for the direction of currents and waves. The azimuth convention is used for the current. This means that0° corresponds to a current going to the positive y direction and90° to a current going to the positive x direction. For the

(34)

Chapter 2. Theoretical background 18

Figure 2.4: Wave input for MoorDyn-UGent.

wave input the meteorological convention is handled. 0° coincides with a wave coming from the positive y direction and90° means that the wave is coming from the positive x direction. The options for the post-processing of the results are defined in a separate text file. Figure 2.5 shows an example of the post-processing options.

Several tools are available to visualize the results. In this project Python scripts and the program ParaView are used to look at the behaviour of the mussel line.

For further information on how to run a simulation in MoorDyn-UGent the reader is referred to the User’s Guide by Pribadi and Donatini (Pribadi and Donatini, 2018b).

2.2 Experimental modelling

The experiments in this project were performed with a scale model of a semi-submerged long-line. Froude scaling was used to establish the scaling factors between the model and the full-scale set-up. A geometric scaling ratio is defined first (Chakrabarti, 1994):

𝜆 = 𝐿𝐿𝐹𝑆

𝑀 (2.4)

in which FS stands for full-scale and M denotes model. In the present case the water depth was used to determine𝜆. The real water depth was 30 m and the depth of the tank was 1.5 m, so 𝜆 was set to 20. All other scaling factors are derived from the geometric scaling ratio.

A Froude number is given by:

𝐹𝑟 = 𝑈√

(35)

2.2. Experimental modelling 19

(36)

Chapter 2. Theoretical background 20 with U a characteristic velocity, g the standard gravity and L a characteristic length. Applying Froude scaling means matching Fr of model and full-scale. This leads to:

𝑈𝐹𝑆 = 𝑈𝑀

√ 𝐿𝐹𝑆

𝐿𝑀 (2.6)

The velocity scaling factor is thus√𝜆. Methods to determine scaling ratios for other parameters are similar and can be found in Chakrabarti (1994). The scaling ratios of the most important parameters and their value in this project are depicted in table 2.1.

Table 2.1: Scaling factors according to Froude scaling.

Variable Units Scaling ratio Value

Length m 𝜆 20 Velocity m/s √𝜆 4.47 Acceleration m/s2 1 20 Volume m3 𝜆3 8000 Mass kg 𝜆3 8000 Mass/length kg/m 𝜆2 400 Force kg m/s2 𝜆3 8000 Young’s Modulus kg/(m s2) 𝜆 20 Time s √𝜆 4.47

These values were used as a guide to choose materials and build the model. On many occasions the real values differ from these preliminary numbers, as the exact scale is not necessary. The goal was to recreate the built model in the MoorDyn software.

(37)

Chapter 3

Methodology

3.1 Model tests

Regular waves of differing amplitude and frequency were applied to the scale model of a semi-submerged mussel line. In general wave heights were increased with a step of5 cm from 10 to 30 cm. Wave periods ranged from 0.5 s to 2.5 s with a step of 0.5 s. Certain combinations were not possible: the wave maker didn’t start or halted midway in the test. Other combinations did not result in a sinusoidal wave. These results were not used in the validation. If a wave period of0.5 s was not possible, a period of 0.7 s was chosen. If the latter was still not possible, a first wave period of0.8 s was opted. The tests were performed for four incoming wave angles, namely 0, 30, 60 and 90°. 0° corresponds with the physical situation of waves parallel to the longline. Every test was run for 60 s to allow the system to reach a periodic behaviour. A waiting period between all tests allowed the flow to settle and prevented any influence from a previous test. Underwater videos to see the behaviour of the longline were recorded using a GARMIN VIRB X.

3.1.1 MC-Lab

The model tests were performed at the Marine Cybernetics Laboratory (MC-Lab) at NTNU (Norwegian University of Science and Technology) in Trondheim (https://www.ntnu.edu/ imt/lab/cybernetics).

The MC-Lab is a small wave basin with tank dimensions L x B x D =40 m x 6.45 m x 1.5 m. The water level in the tank was measured multiple times and had a constant value of1.47 m. The

(38)

Chapter 3. Methodology 22 MC-Lab uses a right-handed inertial reference system. The x-axis lies along the length of the tank with the same direction as the travelling waves. The z-coordinates are considered positive upwards. The same reference system is used in the MoorDyn-UGent model.

The test section of the wave basin is shown in figure 3.1. Two movable bridges can be seen in the picture. The model was put into the water from the first bridge. The second bridge, also usable as towing carriage, contains the processing tools.

Figure 3.1: Test section of the wave basin in the MC-Lab.

In this section the available systems in the MC-Lab will be described. The working principle of the wave maker is illustrated first. The following sections handle the position and force measurement facilities. Finally, the manner of post-processing of the data is described.

(39)

3.1. Model tests 23

Wave maker

The wave maker is a single paddle wave generator with electrical servo actuation, delivered by DHI. The in-house MCLSkvulp V6.0 software is used to control the paddle. Regular waves with a wave height below0.30 m and periods between 0.3 and 3 s can be generated. Several irregular wave spectra are available with a significant wave height below0.15 m and periods between 0.6 and1.5 s.

There were some discrepancies between the wave characteristics that were chosen as input and the wave that was actually created. In general the amplitude of the physical wave was several centimetres smaller than the expected wave. This was especially true for the higher amplitudes. The wave maker also experienced difficulties generating the higher frequencies, corresponding to wave periods of0.5, 0.7 and 0.8s. A possible explanation is the increasing mechanical friction due to wear of the system over the years. A capacitance wave gauge was added to the set-up to measure the physical wave. The measured wave was then used as MoorDyn-UGent input. A damping beach is situated at the opposite end of the wave maker to absorb the waves and limit reflection.

Position measurement

The position of the buoy was recorded using Oqus cameras and the Qualisys Track Manager (QTM) software, both by Qualisys. This motion capture technology operates as follows. Re-flective balls are mounted on the model first. These balls reflect the infrared waves emitted by the system. Three high speed infrared cameras capture the reflected radiation and the exact position of a ball is calculated with triangulation by the QTM software. Figure 3.2 shows the arrangement of the Oqus cameras on the second bridge. The recorded position data is then transferred to the processing computer with a sampling frequency of50 Hz.

During trials a significant pitch motion of the buoys was noted. Therefore, to know the exact position of the bottom of the buoy all six degrees of freedom had to be measured. To this end three reflective balls were mounted on top of each buoy. The exact position of the bottom of the buoy relative to the balls was introduced into the software. The positioning system then measured the position of the balls and automatically calculated the location of the bottom of the buoy. The mass of the three balls together was40 g.

(40)

Chapter 3. Methodology 24

Figure 3.2: Position of the Oqus cameras used for motion capture.

Force measurement

The tension force in the chains was measured using submersible load cells based on regular strain gauges. Four force sensors were used in total. A sensor replaced one shackle in each chain close to the buoy and the anchor. At some point one sensor was no longer usable when water intruded into the coating. It was replaced with a spare one. The final tests were performed with three working sensors since no new ones were on hand. The sampling frequency was200 Hz. The complete data sheets of these force sensors can be found in Appendix A.

As can be seen in the data sheets, the force sensors used are more suited to measure larger forces than those encountered in this project (100 N versus maximum measured forces of about 1 N). These force sensors were the smallest ones on hand.

Data was transferred from the model to the signal processing unit with wires. Care was taken that these wires were never taut or wrapped around the chains, as to minimise their effect on the motion of the model.

An additional force sensor check was performed on every sensor used. This was done to check and possibly adapt the calibration. Several weights, ranging from50 g to 2.5 kg, were hung from the sensor. The corresponding forces were noted and a best-fit curve was constructed. The new offset and slope were introduced into the software.

(41)

3.1. Model tests 25 The real-time sensor values showed the habit of drifting when left alone. Therefore before each test a zero-averaging was executed. Values were measured during ten seconds and the average of these values was chosen as the new zero point. During the tests the deviation with respect to this starting value was measured and not the physical force. The knowledge of the pretension in the chain in calm water was needed to calculate the total physical force. The total tension force is then given by the summation of a static and a dynamic component (the pretension and the force due to wave action, respectively):

𝑇𝑡𝑜𝑡 = 𝑇𝑝𝑟𝑒𝑡𝑒𝑛𝑠𝑖𝑜𝑛+ 𝑇𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 (3.1)

The pretension force was calculated with a catenary equation according to Jonkman and Buhl (Jonkman and Buhl, 2007). This method also accounts for line elasticity and seabed interaction, next to the weight of the chain. The values according to this method amounted to 0.9674 N and1.588 N at the location of the anchor and buoy sensors, respectively. A MoorDyn-UGent simulation without waves was also performed to assess the pretension force. The simulated pretensions at the location of the anchor and buoy sensor were0.9225 N and 1.4125 N, respec-tively. These values make up the largest part of the forces present in the system, since values over0.6 N were seldom measured. To be able to make a good comparison between the measured and simulated forces due to wave action, the pretension predicted by MoorDyn-UGent is used to calculate the total tension force in the experiments.

It should be noted that this force calculation is an approximation. The pretension depends on the position of the buoys. As the position of the buoys alters during the experiments, so does the static pretension load in the chain.

Data processing

The QuantumX data acquisition system (DAQ) by HBM was used for the data collection. All sensor data on forces and wave elevation was sent to one amplifier. The position and force measurements and wave elevation were then post-processed and analysed with the catman software by HBM.

The results of the force measurements showed a large amount of noise. A Fourier transfor-mation was performed to find the dominant frequencies. The highest noise levels were found around the net frequency of50 Hz and its multiples. A Butterworth low-pass filter with a cut-off frequency of20 Hz was applied to remove this noise.

(42)

Chapter 3. Methodology 26

3.1.2 Model

The dimensions and layout of the mussel line model are discussed in this part. For every com-ponent, the material and relevant properties are listed.

A downscaled model of an existing set-up is needed to perform the experiments. The chosen set-up is the Edulis test line in the Belwind concession zone. This set-up is illustrated in figure 3.3. The water depth at the location is30 m. The most important line and buoy properties are given in tables 3.1 and 3.2.

Figure 3.3: Edulis test line in Belwind. (Pribadi et al., 2019)

Table 3.1: Line properties of the Belwind test line. (Pribadi et al., 2019)

Line Type [-] Mass per Length [𝐤𝐠/𝐦] Diameter [𝐦] Line Length [𝐦]

Chain (Grade 3 steel) 10.910 0.022 108

Backbone (Movline Plus 8 strands) 2.1 0.068 57

Mussel sock (fully grown mussels) 21.8 0.15 145

Table 3.2: Buoy properties of the Belwind test line. (Pribadi et al., 2019)

Buoy Type [-] Mass [𝐤𝐠] Diameter [𝐦] Length [𝐦] Volume [𝐦𝟑]

SPAR buoy 2500 0.790 8.865 4.345

(43)

3.1. Model tests 27 with the main dimensions. Some differences between both set-ups can be observed immediately. Unlike the full-scale set-up, the model only has SPAR buoys at the end of the backbone. This is sufficient due to the limited length of the model. The model only contains gravity anchors and has no Danforth anchor as a backup. The final difference lies in the shape of the collector lines. Vertical dropper lines are chosen instead of the V-shaped collectors.

Figure 3.4: CAD drawing of the model (lengths given in mm).

An underwater picture of the longline model is shown in figure 3.5.

The anchors are placed5.2 m apart. They were assumed fixed in this study. Two heavy blocks of lead of2 kg were used to make sure they didn’t move during tests. Cable strips were used to fasten the chains to the anchors. An anchor, the chain connection and a force sensor are depicted in figure 3.6. The blue polypropylene rope was applied for easy handling of the anchor. Each chain has a length of 2 m from the connection point at the anchor to the bottom of the buoy. Chains of Grade 3 steel were used in the model. They have a Young’s modulus of193 GPa, a nominal diameter of2 mm and a mass density of 0.065 kg/m.

A rope of in PES spun polyester represents the backbone. The rope has a diameter of4 mm and a mass density of0.012 kg/m. The Young’s modulus E was unknown and had to be determined with a tensile test. The test set-up is shown in figure 3.7. A large water container was hung from the ceiling using the rope. Water was gradually added in steps of1 l and the corresponding elongation was noted.

(44)

Chapter 3. Methodology 28

Figure 3.5: Deployed model ready for testing.

𝜎 = 𝐹𝐴 (3.2)

𝜖 = 𝛥𝐿

0 (3.3)

𝐹 being the calculated weight, 𝐴 the cross-section, 𝛥 the elongation and 𝐿0the original length

of the test section. The resulting stress-strain curve is depicted in figure 3.8. The slope of the trendline fitting the data corresponds with the Young’s modulus. The Young’s modulus is set to2.23 GPa.

19 droppers are fastened to the backbone with swivels. The spacing between two droppers is 10 cm and each dropper has a length of 30 cm. The swivels are pinched to the backbone to fix their position. The droppers can still rotate freely. Swivels and droppers are connected using heat shrinks. The connection of one dropper line is shown in figure 3.9.

(45)

3.1. Model tests 29

Figure 3.6: Anchor and force sensor.

Figure 3.7: Tensile test set-up.

droppers with fully grown mussels. The loads on the system are expected to be maximal at this moment of the culture cycle.

(46)

Chapter 3. Methodology 30

Figure 3.8: Stress-strain curve of the backbone material.

Figure 3.9: Connection of a dropper to the backbone.

height of the buoys is 460 mm. Metal lids, provided with O-rings for sealing, were added to both ends of the tube. The top lid was later removed to lower the centre of gravity. Additional weights were added inside the buoys to keep them stable in the water. The total mass of buoy,

Afbeelding

Figure 2: Measured and simulated forces for waves with a wave period of 1.5 s and incoming wave angle of 0°.
Figure 7: Measured and simulated forces for waves with a wave period of 1 s and incoming wave angle of 90°.
Figure 1.4: Location of wind farms in the Belgian part of the North Sea. (IMC, 2012).
Figure 3.1: Test section of the wave basin in the MC-Lab.
+7

Referenties

GERELATEERDE DOCUMENTEN

 Aangezien er zowel bedrijven zijn in de landbouwregeling als in de BTW-regeling, zijn alle kosten omgerekend naar en weergegeven exclusief BTW; vervolgens is de kostprijs

Thus, this research includes gender as the controlling variable to enrich the results by considering about the influence of rating valence variance closely from

BCI’s zijn ook omslachtig omdat ze alleen werken als van tevoren is bepaald welke hersenactiviteit gekoppeld kan worden aan het besturen van de computer.. In het fMRI-voorbeeld

Se pensiamo che in un articolo di Giuseppe Mazzocchi (A proposito della nuova gramamtica di Manuel Carrera Diáz, p.115) sulla nuova grammatica spagnola di Manúel

The study of the density separated samples allows the gathering of information on mineral distribution in the coal particles and its degree of association with the

De resultaten worden gebruikt om de Pest Risk Analysis af te maken, waardoor de praktijk meer duidelijkheid krijgt over de risico’s van deze nieuwe nematodensoort voor

Deze duiding sluit aan bij de feitelijke situatie waarbij de radioloog de foto beoordeelt en interpreteert, en lost een aantal praktische knelpunten op.. Omdat de

62 Appendix A1: Rankine cycle EES model with 33°C condenser operating temperature Appendix A2: Rankine cycle EES model with 50°C condenser operating temperature Appendix A3: