Optimization and Evaluation of Subsea Cooling for the Offshore Wind Industry A Combined CFD and Market Evaluation for a Passive Subsea Cooler Master’s thesis in MPSES and MPQOM ELSA TÄCK & JONATHAN BLOMBERG DEPARTMENT OF MECHANICS AND MARITIME SCIENCES CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2024 www.chalmers.se www.chalmers.se Master’s thesis 2024 Optimization and Evaluation of Subsea Cooling for the Offshore Wind Industry A Combined CFD and Market Evaluation for a Passive Subsea Cooler ELSA TÄCK JONATHAN BLOMBERG Department of Mechanics and Maritime Sciences Chalmers University of Technology Gothenburg, Sweden 2024 Optimization and Evaluation of Subsea Cooling for the Offshore Wind Industry A Combined CFD and Market Evaluation for a Passive Subsea Cooler © Jonathan Blomberg, 2024. © Elsa Täck, 2024. Supervisors: Andreas Öberg, Senior Specialist CFD and Processing Engineer & Jonathan Haglund, Materials Engineer Examiner: Henrik Ström, Department of Mechanics and Maritime Sciences Master’s Thesis 2024 Department of Mechanics and Maritime Sciences Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: Rendering of Subsea Cooler from CFD Simulations. Typeset in LATEX, template by Kyriaki Antoniadou-Plytaria Printed by Chalmers Reproservice Gothenburg, Sweden 2024 iv Optimization and Evaluation of Subsea Cooling for the Offshore Wind Industry A Combined CFD and Market Evaluation for a Passive Subsea Cooler Jonathan Blomberg & Elsa Täck Department of Mechanics and Maritime Sciences Chalmers University of Technology Abstract With an expanding energy market and an increasing focus on developing green, reli- able and sustainable energy, the demand has never been greater for new innovations meeting these criteria. Today, offshore wind stands as a front-runner amongst the renewable energy alternatives. For the technology to be truly competitive against fossil fuels, the method of transporting large amounts of generated power long dis- tances must be improved. High Voltage Direct Current (HVDC) has proven to be a promising technology offering low transmission losses over greater distances. How- ever, the conversion process from Alternating Current (AC) to Direct Current (DC) which takes place on offshore HVDC platforms is generating a significant amount of heat, resulting in a high cooling demand. Conventional cooling systems are energy consuming and utilize chemicals in the cooling process which are discharged to sea. This thesis has investigated an alternative cooling solution, utilizing subsea cool- ing and a closed loop. The alternative cooling solutions is free of chemicals and was found to be far more energy-efficient in terms of operating cost compared to conventional cooling. The thesis also screened the future market for HVDC plat- forms in the North Sea and its cooling demand, by conducting a progressive market screening. It was concluded that the market is set to triple in terms of amount of HVDC platforms to be established and seven-fold in terms of installed converting capacity over the next seven years. A passive subsea cooler, operating in a closed loop utilizes natural convection and has a unique design to withstand shallow water conditions and lower coolant temperatures. Computational fluid dynamics (CFD) simulations have been performed to evaluate different cases of pipe arrangement and chimney designs to investigate the performance of the proposed subsea cooler. Manufacturing possibilities and their implications were also investigated together with a techno-economic discussion which presents various design alternatives. The study has concluded that, compared to existing technologies in the market, the closed loop cooling system decreases overall operating energy consumption and removes chemical pollutants that previously have been the base case for cooling offshore HVDC platforms. CFD simulations were proven to be an important tool in testing the passive subsea cooler performance. It was identified that baffles and large temperature differences are crucial in enhancing a passive subsea cooler. The thesis also managed to reduce computational time for the full 3D simulation by 96%, to test the chimney design. Keywords: offshore wind, passive cooler, high voltage direct current, offshore plat- form, computational fluid dynamics, market screening, subsea cooling v Acknowledgements We would like to acknowledge the company for giving us the opportunity to investi- gate such an interesting topic. Thank you to our supervisors Jonathan Haglund and Andreas Öberg for always being available and supporting in providing knowledge throughout the progress of the thesis. We would like to express our gratitude to our research examiner Henrik Ström, Professor at the Division of Fluid Dynamics at the Department of Mechanics and Maritime Sciences at Chalmers University of Technology, for taking on this MSc. thesis. And whose expertise, guidance, and encouragement were invaluable through- out the research and writing process. We would also expressed our gratitude towards Max Bergström, Cooler Product Lead at the company, for providing input regarding general knowledge about pas- sive coolers and showing great interest for the thesis. We also wish to thank Anton Riström, CFD Department Manager at the com- pany, for his believing in us and making the thesis possible at the company. Jonathan Blomberg, Gothenburg, May 2024 Elsa Täck, Gothenburg, May 2024 vii List of Acronyms Below is the list of acronyms that have been used throughout this thesis listed in alphabetical order: AC Alternating Current CAPEX Capital Expenditure CFD Computational Fluid Dynamics CuNi Copper Nickel DC Direct Current ESG Environmental, Social and Governance HVAC High Voltage Alternating Current HVDC High Voltage Direct Current ID Inner Diameter OD Outer Diameter OHTC Overall Heat Transfer Coefficient OPEX Operating Expense OWF Offshore Wind Farm SWLP Sea Water Lift Pump ix Contents List of Acronyms ix List of Figures xv List of Figures xv List of Tables xvii List of Tables xvii 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.1 Societal, Ethical, and Ecological Aspects . . . . . . . . . . . . 2 1.1.2 CFD Background . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3.1 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3.2 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . 4 1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 A Review of Current HVDC Platforms & Its Cooling System 7 2.1 Offshore HVDC Platforms . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1.1 Operational HVDC Platforms . . . . . . . . . . . . . . . . . . 7 2.2 Cooling Systems for Offshore HVDC Platforms . . . . . . . . . . . . 9 2.2.1 Current Cooling Systems . . . . . . . . . . . . . . . . . . . . . 9 2.2.2 Subsea Cooling System . . . . . . . . . . . . . . . . . . . . . . 11 2.2.3 Alternative Cooling Systems . . . . . . . . . . . . . . . . . . . 12 2.3 The Subsea Cooler . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3 Theory 15 3.1 Heat Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.2 Heat Exchanger Design . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.2.1 Pipe Arrangements . . . . . . . . . . . . . . . . . . . . . . . . 17 3.2.2 Chimney . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.3 Physics Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.3.1 EB k-ε . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.3.2 SST k-ω . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 xi Contents 3.4 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.5 Important Considerations . . . . . . . . . . . . . . . . . . . . . . . . 22 3.5.1 Solvers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.5.2 Natural Convection . . . . . . . . . . . . . . . . . . . . . . . . 23 3.5.3 Mesh Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 4 Methodology 25 4.1 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.3 Data Collection and Data Processing . . . . . . . . . . . . . . . . . . 26 4.4 Market Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.4.1 Cost Analysis Comparing HVAC and HVDC . . . . . . . . . . 28 4.5 Cooling System Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.6 Design Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4.6.1 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4.6.2 Cooling water Loop . . . . . . . . . . . . . . . . . . . . . . . . 30 4.6.3 Seawater properties . . . . . . . . . . . . . . . . . . . . . . . . 31 4.7 Building the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.8 Meshing the model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.9 Simulation Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.9.1 Pipe arrangements . . . . . . . . . . . . . . . . . . . . . . . . 34 4.9.2 Full 3D model . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.9.3 Model Simplifications . . . . . . . . . . . . . . . . . . . . . . . 36 4.9.3.1 2D model . . . . . . . . . . . . . . . . . . . . . . . . 37 4.9.3.2 Wedge model . . . . . . . . . . . . . . . . . . . . . . 38 4.10 Final Model Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.10.1 Chimney Design . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.10.2 Design Improvements . . . . . . . . . . . . . . . . . . . . . . . 40 4.10.3 Post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.11 Manufacturing Possibilities . . . . . . . . . . . . . . . . . . . . . . . . 42 5 Results & Discussion 43 5.1 Market Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.1.1 Cost Analysis Comparing HVDC and HVAC . . . . . . . . . . 43 5.1.2 CAPEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 5.1.3 OPEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 5.1.4 Total Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.2 Identified HVDC Platforms . . . . . . . . . . . . . . . . . . . . . . . 48 5.3 Cooling System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 5.3.1 Sea Water Lift Cooling . . . . . . . . . . . . . . . . . . . . . . 50 5.3.1.1 Sea Water Lift Pump . . . . . . . . . . . . . . . . . . 50 5.3.1.2 Electrochlorination Unit . . . . . . . . . . . . . . . . 51 5.3.2 Subsea Cooling . . . . . . . . . . . . . . . . . . . . . . . . . . 51 5.3.3 Comparing the Cooling Systems . . . . . . . . . . . . . . . . . 52 5.4 Mesh Refinement Study . . . . . . . . . . . . . . . . . . . . . . . . . 53 5.4.1 Mesh Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 5.5 Initial CFD Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 xii Contents 5.5.1 Pipe Arrangements . . . . . . . . . . . . . . . . . . . . . . . . 56 5.5.2 3D model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.6 Model Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 5.6.1 3D model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 5.6.2 2D model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 5.6.3 Wedge model . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 5.6.4 Computational models discussion . . . . . . . . . . . . . . . . 62 5.7 Final design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 5.7.1 Chimney design . . . . . . . . . . . . . . . . . . . . . . . . . . 63 5.7.1.1 Chimney Design Discussion . . . . . . . . . . . . . . 64 5.7.2 Design Improvements . . . . . . . . . . . . . . . . . . . . . . . 64 5.7.2.1 Design Improvements Discussion . . . . . . . . . . . 67 5.7.3 Final design Discussion . . . . . . . . . . . . . . . . . . . . . . 67 5.7.3.1 Further CFD Work Discussion . . . . . . . . . . . . 68 5.8 Manufacturing Possibilities . . . . . . . . . . . . . . . . . . . . . . . . 69 5.8.1 Piping Bundle . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 5.8.2 Chimney . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.8.3 Techno-Economic Discussion . . . . . . . . . . . . . . . . . . . 71 6 Conclusion 73 6.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 6.2 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 6.3 Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Bibliography 77 xiii Contents xiv List of Figures 2.1 Reference image of HVDC platform [1] . . . . . . . . . . . . . . . . . 8 2.2 Conventional cooling loop . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3 Subsea cooling loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.4 Exploded view of subsea cooler . . . . . . . . . . . . . . . . . . . . . 13 3.1 Reference passive cooler . . . . . . . . . . . . . . . . . . . . . . . . . 17 3.2 Example of tube arrangement parameters . . . . . . . . . . . . . . . . 18 3.3 Example of the chimney set-up by Kumar [2] . . . . . . . . . . . . . . 19 4.1 Snapshot of TGS 4C offshore interactive map . . . . . . . . . . . . . 27 4.2 Subsea cooler domain setup . . . . . . . . . . . . . . . . . . . . . . . 32 4.3 Pipe arrangement seawater set-up . . . . . . . . . . . . . . . . . . . . 34 4.4 Pipe arrangement designs . . . . . . . . . . . . . . . . . . . . . . . . 35 4.5 Boundary conditions on the cooler . . . . . . . . . . . . . . . . . . . . 36 4.6 Temperature profiles from simulation wall . . . . . . . . . . . . . . . 37 4.7 2D model surface naming matrix . . . . . . . . . . . . . . . . . . . . 38 4.8 Temperature distribution on pipe surfaces . . . . . . . . . . . . . . . 39 4.9 Wedge tests designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 5.1 Investment cost compared to total cable length . . . . . . . . . . . . 45 5.2 Operating expenses compared to cable length . . . . . . . . . . . . . 46 5.3 Total cost over lifetime compared to cable length . . . . . . . . . . . 47 5.4 Identified HVDC projects and average converting capacity . . . . . . 49 5.5 Mesh scenes of the seawater domain and of the refinements of the pipes 54 5.6 Mesh refinement scene of the surface mesh for 8mm and 6mm Surface size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.7 Scenes of velocity streamlines for a) original pipe arrangement b) additional spirals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.8 Scenes of velocity streamlines for a) smaller SL b) larger SL . . . . . . 57 5.9 3D Scenes of velocity streamlines for a) Wall b) Chimney c) Baffles . 59 5.10 2D simulations scenes of velocity streamlines for a) Wall b) Chimney c) Baffles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 5.11 Wedge simulations scenes of velocity streamlines for a) Wall b) Chim- ney c) Baffles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 5.12 Wedge simulations scenes of velocity streamlines for all Chimney Sims 63 5.13 Simulations scenes of temperature profile for BafflesT45 . . . . . . . . 64 5.14 Simulations scenes of velocity streamlines for No Wall . . . . . . . . . 65 xv List of Figures 5.15 Simulations scenes of velocity streamlines for a) Baffles b) BafflesT45 66 5.16 Simulations scenes of velocity streamlines for a) HTA70 b) HTA70T45 66 xvi List of Tables 2.1 Operational offshore HVDC platforms [3] . . . . . . . . . . . . . . . . 9 4.1 Pipe material specification . . . . . . . . . . . . . . . . . . . . . . . . 30 4.2 Cooling fluid specification . . . . . . . . . . . . . . . . . . . . . . . . 31 4.3 Seawater specification . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.4 Automated mesh custom controls . . . . . . . . . . . . . . . . . . . . 33 5.1 Cost data from bidding contracts for established HVDC platforms . . 44 5.2 Average PoE North Sea countries . . . . . . . . . . . . . . . . . . . . 46 5.3 OPEX per km in range 100 to 200 km . . . . . . . . . . . . . . . . . 47 5.4 Input data for sea water lift pump calculations . . . . . . . . . . . . . 50 5.5 Input data for cooling medium pump . . . . . . . . . . . . . . . . . . 52 5.6 Mesh settings for refinement study . . . . . . . . . . . . . . . . . . . 53 5.7 Mesh refinement result . . . . . . . . . . . . . . . . . . . . . . . . . . 55 5.8 Mesh study computational time results . . . . . . . . . . . . . . . . . 55 5.9 Pipe arrangement results . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.10 3D model results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.11 3D model validation results . . . . . . . . . . . . . . . . . . . . . . . 58 5.12 2D model comparative results . . . . . . . . . . . . . . . . . . . . . . 59 5.13 Wedge model comparative results . . . . . . . . . . . . . . . . . . . . 61 5.14 Computational time for 3D and wedge model of simulation "Baffles" . 62 5.15 Chimney design results . . . . . . . . . . . . . . . . . . . . . . . . . . 63 5.16 Design improvement results . . . . . . . . . . . . . . . . . . . . . . . 65 xvii List of Tables xviii 1 Introduction This chapter describes the background to the issue being investigated and how CFD can be applied. Additionally the problem is described, together with the purpose, aim, limitations and research questions the thesis aims to answer. 1.1 Background In a time characterized of climate change and increasing green house gas emissions, the need for green, reliable and renewable energy has never been more significant. COP28 in Dubai concluded the first global assessment under the Paris Agreement which measured the progress towards the climate goals set by the agreement. Re- newable energy and energy efficiency was highlighted as the most important aspects to mitigate climate change and it was agreed amongst the party members to triple global renewable energy capacity and double the rate of energy efficiency improve- ments by 2030 [4]. Among the renewable energy alternatives available today, offshore wind stands as a front-runner offering almost limitless opportunities. The Interna- tional Energy Agency predicts that the global offshore wind industry is set to 15-fold its capacity over the next two decades turning it into a 1 trillion dollar industry, with potential for further growth as climate targets are becoming more ambitious, making the industry even more attractive for investors [5]. Offshore wind could be deployed in large scale without having an impact on land use, meaning that offshore wind turbines could be much larger, than onshore ones, allowing for an increase in energy output per wind turbine produced. The wind speeds offshore are generally higher and more reliable, offering a mitigated intermittence of the energy produc- tion, which currently is a challenge for renewable energy production. Currently, offshore wind is expensive to build and operate, resulting in slimmer margins than for instance fossil fuels, such as oil and gas [6]. For offshore wind to be competitive against alternative energy sources, such as oil and gas it has to cope with transporting electricity long distance and avoiding severe energy losses in the process. Offshore wind farms (OWF) today are located more than 70 kilo- metres from shore and generate large quantities of alternating current (AC) power through its wind turbines. Transporting AC power such distances generates signif- icant losses, resulting in a need to convert the electricity into direct current (DC) and ramp up the voltage before being transmitted to shore. The conversion process takes place on a high voltage direct current (HVDC) offshore platform located in close connection to the OWF and becomes significant as the positioning of OWF 1 1. Introduction are becoming greater from shore [7]. When the electricity has reached shore, it is converted back to AC to be used by consumers. The conversion process on the plat- forms generates a byproduct of heat that has to be removed in order to not damage the conversion and transformation equipment. Since the platform is placed offshore in harsh environment the equipment has to be enclosed on the platform as to avoid saltwater damage which makes cooling by ambient air difficult. The function of the cooling system is therefore critical in order for the OWF to be operational. A failure in the cooling system would result in a shutdown of several wind turbines or the entire OWF, which is costly both in terms of lost revenue due to decreased production and excessive cost of sending service personnel to the platform. The only commercially available technology today to cool these platforms is by an open loop system which utilize seawater for cooling. This technology have a rather high operating cost, requiring electricity that has to be scavenged from the OWF. The system also uses chemicals that are discharge to sea which has proven to have an environmental impact. As an alternative to the open loop system a closed loop system utilising cooling through natural convection has been investigated. A new type of subsea cooler has therefore entered the preliminary design stages to serve as a passive cooler in a closed loop system, which is further investigated in this thesis. 1.1.1 Societal, Ethical, and Ecological Aspects The development and implementation of a new subsea cooler used for offshore wind, will have an impact on the society as well as the environment. Offshore wind is a crucial energy source that will play a major role in the green transition. Innovative solutions that allow renewable energy production to stay competitive against fossil fuels is of great importance in order to achieve the green transition. Furthermore, the deployment of an innovative subsea cooler for offshore wind will not only represent a leap forward in the green transition but also offer the benefit of new jobs being created in local communities and economic development. 1.1.2 CFD Background Computational Fluid Dynamics (CFD) is a field of fluid mechanics that uses nu- merical analysis to solve and analyze problems involving fluid flows. By employing powerful computational techniques, CFD allows engineers to simulate the behavior of fluids under various conditions, providing detailed insights into the interactions between fluids and surfaces [8]. CFD is particularly valuable tool for evaluating the performance of a subsea cooler and has been used by company in previous subsea cooler projects. It enables visu- alization and analysis of fluid flow patterns around and through the cooler, helping to identify areas of high resistance that can impact efficiency. Additionally, CFD simulations model the thermal performance of coolers by analyzing heat transfer processes, allowing for predicted temperature distributions and ensure optimal op- erating temperatures. Moreover, CFD helps in optimizing the cooler’s design by simulating different configurations and operating conditions. Different geometries 2 1. Introduction and materials could be tested virtually, leading to improved designs that maxi- mize cooling efficiency in harsh offshore environments without extensive physical prototyping. Traditional testing and prototyping methods are often expensive and time-consuming, especially for subsea equipment. CFD reduces the need for physi- cal prototypes by providing accurate virtual simulations, significantly cutting down on development time and costs. By predicting potential performance issues and failure points, CFD also helps in identifying and mitigating risks early in the design process, enhancing the reliability and safety of subsea coolers [8]. 1.2 Problem Description With major energy companies rapidly shifting their focus from fossil fuels, which have been the main energy source for decades, to green and renewable energy, there is a growing demand for new innovations to drive the energy market forward. Subsea coolers has previously been developed for oil and gas applications which are placed in much deeper depths than offshore wind which requires much stricter standards, adding additional cost to the cooler. Furthermore, the cost of offshore wind is com- paratively high and therefore all of the components of an OWF has to be evaluated in order for it to be a competitive energy alternative. Since the subsea cooler and its associated system is in a concept phase, the actual performance of the cooler and the system is still not determined. This also applies to the manufacturing aspect of the cooler where there also is a need for further investigating. In order to make the technology commercially viable, the concept has to be proven and evidence of why this technology is a better alternative have to provided. 1.3 Purpose The overall purpose of the thesis is to have a dual perspective. Meaning that both commercial aspects such as market analysis and manufacturing possibilities and its cost implications is investigated together with the technical aspect that CFD is offering by investigating the cooler performance. The final aim is then to merge the findings to provide a conclusion whether development of such a subsea cooler utilising a closed loop cooling system is worth it or not. The aim is that the conclusion is based on both perspective, meaning that it rests on both the market outlook, the associated cost savings with a closed loop system and how the design of the cooler performs. To arrive at this conclusion the thesis aim to use CFD to try different designs and evaluate how that effects parameters such as, outlet temperature, heat transfer etc. Additionally the thesis aims to conduct a technical market analysis to arrive at an estimated market outlook for HVDC platforms and its cooling demand and the cost implications of a subsea cooling system compared to a conventional cooling system. Finally, the aim is to investigate the manufacturing possibilities from a cost perspective. 3 1. Introduction 1.3.1 Limitations The thesis will focus on the cooler being used for offshore wind applications requiring offshore HVDC platforms, meaning that the inlet and outlet temperatures temper- ature are determined with this application in mind. Regarding the market analysis it is limited to only focus on the North Sea. Regarding the cooling system, it is currently in a concept stage meaning that data are used for similar application and not specifically for HVDC platforms. Additionally, a limitation is the intermittency of wind power, since the production is not continuous, electricity prices can vary significantly between full load hours based on demand from consumers, which is not considered in this thesis, were average electricty prices are used. For the CFD analysis any structural components will not be included in the sim- ulations in order to reduce computational time. Another limitation of the thesis the OHTC is based on the mesh area, which depends on the surface refinement and amounts of points on the circle, which means the area is smaller than in reality. Ar- ticles found and investigated mostly discuss heat exchangers with air on the outside of the tubes, which deviates from the thesis cooler. 1.3.2 Research Questions How can CFD facilitate the optimization of a new design of subsea cooler, and how can different design aspects affect the performance? • To what degree can the testing and evaluation of a subsea cooler model be improved by simplifying its features to reduce computational time in CFD simulations? • How can a subsea cooler be optimized when investigating the relationship between design and performance? What are the key market dynamics and factors influencing the successful entry of the new subsea cooler in the wind power market, and how can strategic marketing and positioning be leveraged to ensure a sustainable competitive advantage in the evolving market landscape? • To what extent can existing technologies and knowledge be innovatively ap- plied to meet the demands of the new changing market? 1.4 Thesis Outline The thesis contains chapters and sections based on the content required for the reader to have an understanding for the structure of the thesis. Each chapter introduces a new stage in the thesis, and the sections are designed to divide the information into logical order, beginning with the larger market, and successively working towards the technical aspects and the product specifications. The thesis is initiated by a review of current HVDC platforms and its asso- ciated cooling system located in the North Sea and a deeper investigation about alternative cooling systems to and the subsea cooling system in detail, which serves 4 1. Introduction as a foundation for the cost comparison between the systems. The Theory chapter describes the underlying theory for heat transfer and heat exchangers necessary for the CFD analysis and the computational models and equa- tions that are important considerations such as solvers and mesh theory. The Methods chapter is initiated by general considerations for writing a thesis such as a literature review and data collection. This is then followed by the method of the market and cooling system analysis. The design basis of the thesis is described followed by the construction, meshing, and simulating set-up for the CFD analysis, which is concluded by the model design parameters. Finally the costing estimation is described. Results and Discussion chapter similar to the previous chapter is initiated by the market analysis and mapping results, followed by the cooling system analysis results. For the CFD analysis the mesh refinement study is presented, followed by some preliminary CFD results and the model validation. The final cooler design is then presented and discussed together with the product costing and weight estima- tion results. In the Conclusion chapter, the thesis summarize its findings in regards to the market, CFD and manufacturing possibilities. 5 1. Introduction 6 2 A Review of Current HVDC Platforms & Its Cooling System As an initial step of the thesis a review over current offshore HVDC platforms and its associated cooling systems was conducted. The current market is used for comparison between the market screening at a later stage in the thesis. 2.1 Offshore HVDC Platforms The need for HVDC offshore platforms has emerged over the last couple of years, as more and more OWF have been established further from shore where there is an increasing need for transmitting generated electricity longer distances [9]. The devel- opment in technology and increased demand of green energy has made OWFs grow and significantly increase its generated electricity. The biggest advantage presented by HVDC technology lies in its significantly reduced transmission losses compared to HVAC systems. This feature is of great importance, particularly in the trans- portation of substantial quantities of electricity [10]. The wind turbines that makes up the OWF are generating AC, which creates a need to convert the generated AC electricity into HVDC which is more suitable for long transportation of electricity to shore [7]. Hence, offshore HVDC platforms has to be installed in close proximity to the OWF and collect the electricity from several wind turbines were it is con- verted into DC and the voltage is increased. The platforms are connected to large HVDC interconnected subsea export cables that are transmitting the electricity to an Onshore Converter Station (OCS), where it is being converted back into AC to be used by consumers. The platforms are placed in harsh conditions and have to cope with all sorts of weather, which makes it crucial to ensure high standards for the equipment to guarantee reliability. 2.1.1 Operational HVDC Platforms The majority of OWF established today are using HVAC to transmit electricity to onshore consumers. The reason being that wind turbines generate AC and the elec- tricity does not have to converted and could be transmitted directly to shore. The advantage is that the current does not have to be converted to DC, hence there is no need to establish offshore converting infrastructure such as HVDC platforms. The rather high investment cost of building the HVDC infrastructure required is 7 2. A Review of Current HVDC Platforms & Its Cooling System Figure 2.1: Reference image of HVDC platform [1] avoided and many cases that is the cheaper alternative [7]. In recent years however, more and more OWF operators have chosen to opt for DC rather than the AC, which is mainly due to the fact that OWFs are being established further from shore, which means that the electricity has to be transmitted further distances. Today there are nine operational HVDC platforms in the North Sea that are all are owned by the Dutch energy company TenneT. The platforms are serving multiple wind farms in the North Sea outside Denmark’s, The Netherlands’s and Germany’s coast and transmitting HVDC to OCS where the electricity is being converted back to AC and fed into the grid. The OWFs providing power to the HVDC converter platforms are of comparatively smaller scale, typically ranging between 200 to 450 megawatts (MW), in contrast to the larger-scale wind farms which are planned and developed. The distance to OCS are ranging from 130 - 200 km, and the wind farm capacity is for each associated platform are ranging from 400 - 900 MW [3]. The platforms are positioned in the DolWin-, SylWin-, HelWin-, and BorWin Cluster, which is outside Germany’s northwest coast in the North Sea. The HVDC platforms are listed in Table 2.1, where the distance to onshore converter station (DTOCS), ca- pacity and commissioning year is listed. The analysis over current HVDC platforms shows that the the average DTOCS is 165 km and installed conversion capacity is 6,845 GW, which yields an average of approximately 760 MW. The average depth of where these platforms are positioned was also investigated and where found to be approximately between, 40-70 meters. This is however limited by the foundations for anchored wind turbines and in line with the design depth of the cooler. 8 2. A Review of Current HVDC Platforms & Its Cooling System Table 2.1: Operational offshore HVDC platforms [3] HVDC Platform DTOCS (km) Capacity (MW) Commissioned DolWin Alpha 165 800 2015 DolWin Beta 135 916 2016 DolWin Gamma 160 900 2017 SylWin Alpha 205 864 2015 HelWin Alpha 130 575 2015 HelWin Beta 130 690 2015 BorWin Alpha 200 400 2015 BorWin Gamma 160 900 2019 BorWin Beta 200 800 2015 2.2 Cooling Systems for Offshore HVDC Plat- forms The conversion of AC to DC for offshore wind farms is necessary to mitigate energy losses in the transportation process and enables for wind energy to become a cost competitive alternative. In the conversion process a significant amount of heat is generated which has to be drawn away in order to not damage the equipment, which creates a cooling demand [7]. 2.2.1 Current Cooling Systems Current cooling systems on HVDC platforms for offshore wind utilize an open loop cooling solution, where cold water is pumped up to the platform where the cooling process takes place and then discharged back to sea. The cool sea water is passing through a topside heat exchanger positioned topside on the platform where the hot cooling medium is cooled and pumped to the cooling consumers, i.e. converters and transformers on the platform. To maintain a cold stream of sea water to the plat- form a Sea Water Lift Pump (SWLP) is used, which continuously pumps sea water up to the heat exchanger on the the platform. The cooling system is therefore con- sisting of two separated loops connected by the heat exchanger, one open sea water loop and one closed cooling medium loop, located on the platform. The open-loop system is the most effective way of cooling HVDC platforms and the only system in place for operational platforms today. Allowing for the most reliable, effective and cost-efficient solution for AC to DC conversion offshore [7]. In order for the sea water loop to function properly and mitigate required maintenance, the sea water is filtered before it being pumped into the heat exchanger. Coarse filters is placed after the inlet, where sand grains and other marine species, such as plankton or fish is filtered out before reaching the heat exchanger. These filtration systems are often backflushed to decrease maintenance and allow for continuous operation to ensure no unwanted objects are entering the heat exchanger, which may result in a blockage of the sea water loop. When the filtration system is backflushed, some of the marine species caught in the filters are returned back to the ecosystem. However, many 9 2. A Review of Current HVDC Platforms & Its Cooling System Figure 2.2: Conventional cooling loop marine species are not able to mature and reproduce once it has been flushed out of the filtration system [7]. Figure 2.2 is a schematic over the cooling loop and its components. To mitigate the risk of biofouling in the open cooling loop, a dose of sodium hypochlo- rite is injected into all areas where marine biofouling is at risk of occurring and preventing marine growth in exposed areas from starting. The sodium hypochlorite The hypochlorite is produced in the electrochlorination units, by passing electric current through the sea water. To perform the electrolysis of sea water to produce a chlorinated solution, the sea water has to be filtered, which is already done by the filters earlier in the loop. Then the sea water is passing through an electrolyzer cell’s channel of decreasing thickness, where one side is a cathode and the other an anode. The electrolysis takes place when a low voltage DC is applied producing sodium hypochlorite and hydrogen gas [11]. After the sodium hypochlorite has been used in the cooling loop it is discharged to sea. Equation 2.1 is showing the chemical process of producing sodium hypochlorite through electrochlorination. NaCl + H2O + ENERGY ⇒ H2 + NaOCl (2.1) To control the flow of cold sea water entering the heat exchanger sea water booster pumps are used. In many platforms today three sea water booster pumps are used, where one is a spare for redundancy. The booster pumps are variable speed pumps, meaning that the out put rate could be controlled in order to be aligned with the 10 2. A Review of Current HVDC Platforms & Its Cooling System cooling need on the platform. Since the sea water is variable throughout the year, being warmer in the summer and colder during the winter, the pump rate of sup- plying sea water to heat exchanger has to be adjusted accordingly. During most of the time, only one sea water booster pump is active and the second one is activated only in cases when the sea water is exceptionally warm. The increased flow rate of cold sea water will increase the cooling capacity of the system to cope with warmer sea water temperatures. 2.2.2 Subsea Cooling System A subsea cooling system is a new concepts when it comes to cooling HVDC plat- forms and is utilizing a closed loop rather than an open seawater loop as in the conventional cooling loop. The overall philosophy of the system is to move the en- tire cooling process subsea, by using one subsea cooler. Instead of lifting seawater to the platform the entire cooling process is taking place subsea by utilising the natu- ral convection and the cooler seawater to draw away heat from the cooling medium. Subsea cooling would mean that, several components could be eliminated, which includes, pumps, the electrochlorination unit and the filter which is illustrated in 2.2. The electrochlorination unit is eliminated since there is no risk for biofouling when using a closed loop, since the system is not continuously supplied with fresh seawater. This would mean that no sodium hypochlorite is discharge to sea and a mitigated environmental impact. A subsea cooling system would also free up space on the actual platform, otherwise occupied by the pumps and the electrochlorination unit. The only pump present in the subsea cooling system is the cooling medium circulation pump that would be placed topside on the actual platform. There also no need for an intake caisson nor a discharge caisson that is needed in the con- ventional cooling system. The passive subsea cooler illustrated in Figure 2.3 is the same cooler as described in Figure 2.3 that is placed on the seafloor. The hot cooling medium coming from the cooling consumers would enter through the inlet header, pass through the piping bundle were it is cooled and then be pumped up to the cooling consumers again and closing the loop. 11 2. A Review of Current HVDC Platforms & Its Cooling System Figure 2.3: Subsea cooling loop 2.2.3 Alternative Cooling Systems The corrosive nature offshore limits the alternatives to other cooling solutions than the one described under 2.2.1. The electrical equipment has to be enclosed to stay protected, from salt water and lightning, which limits air cooling capabilities. Com- monly air is used to cool the water in a closed-loop system, by having a large bank of electric powered fans cooling the water. When taking the cooling requirements and limited space on the platform into account for such a solution, it would be hard to motivate an air cooled solution over an open looped water cooled. Having several fans is also resulting in an excessive amount of movable parts, which requires contin- uous maintenance and electric consumption that would have to be scavenged from the wind farm, resulting in decreased output from the wind farm to shore. Other cooling systems uses refrigerant gases to draw heat from the water and dissipates it to ambient air. The refrigerant gases is often chlorofluorocarbons and hydrofluoro- carbons that has to replenished over time and the dissipated heat from the water has to be removed by using fans which is power consuming and not suitable to highly corrosive offshore conditions. This solution would also most probably result in a higher overall cost of operating the cooling system [7]. By investigating several different alternatives to cool the conversion process on the HVDC platforms, it was concluded that the current open loop cooling system and the subsea cooling system are the most suitable. 12 2. A Review of Current HVDC Platforms & Its Cooling System 2.3 The Subsea Cooler The passive subsea cooler illustrated in Figure 2.3, is illustrated in more detail by a 3D model provided by company, see Figure 2.4. This initial 3D model is used throughout the thesis and the figure displays the names of each part of the cooler to have consistent terminology usage throughout the thesis. Figure 2.4: Exploded view of subsea cooler Each part of the cooler has been specified below: • Chimney refers to the final part sitting on top of the wall support. This part will be tested in the technical design and manufacturing possibilities will be investigated. • Wall refers to 4 round sheets attached through the wall support. This part will be simulated as one sheet. • Pipe Support will not be investigated in this thesis. • Piping Bundle refers to the 20 pipe stacks that together forms the part transporting the cooling medium through the heat transfer process. This part will be tested in the technical design and manufacturing possibilities will be investigated. • Inlet Header refers to the pipe transporting the cooling medium into each pipe stack. Pumped from the HVDC platform. • Outlet Header refers to the pipe transporting the cooling medium out of each pipe stack. Pumped to the HVDC platform 13 2. A Review of Current HVDC Platforms & Its Cooling System • Bottom Frame will not be investigated in this thesis. An important consideration for the thesis is that the seawater being the cooling medium for this cooler is referred to as seawater. While the fluid inside the pipes being the cooled medium becomes the cooling medium when looking at the applica- tion which is why henceforth the fluid inside the pipes are referred to as the cooling medium or fluid. 14 3 Theory This thesis investigated a subsea cooler with CFD, which requires a lot of background understanding of transport phenomena. This chapter is meant to act as a guide for the thermal, and fluid dynamic problems that is solved, as well as suggesting conventional design requirements for a thermodynamic CFD problem. 3.1 Heat Transfer The two modes of heat transfer relevant for the thesis was conduction and convection. Thermal conductivity is molecular interaction where with higher energy (temper- atures) the greater motion of a molecule affects the energy of adjacent molecules with lower energy levels. Which will be applied to the pipes and specified as a solid physics model. Convective heat transfer occurs with energy exchange between a sur- face and an adjacent fluid [12]. There are two types of convection free and forced, forced convection will be used in the cooling fluid domain, and can be defined as the fluid (at a temperature) flowing in a tube (with a different temperature) at an average velocity imposed by an external agency such as a fan or pump. The relation between conduction and convection can be described in terms of boundary layers in which the closest boundary to a stationary surface the fluid will be laminar, while turbulence will cause bulk mixing of the fluid between regions at different tempera- tures. There are a few significant parameters in convective heat transfer, the Prandtl number being the ratio of the molecular diffusivity of momentum to the molecular diffusivity of heat, and the Nusselt number being the ratio of conductive thermal resistance to the convective thermal resistance of the fluid [12]. In order to evaluate the convective heat transfer coefficient there are four meth- ods; dimensional analysis which require experimental results; exact analysis of the boundary layer; approximate integral analysis of the boundary layer; and analogy between energy and momentum transfer. Generally overall heat transfer is described with temperature difference and through a series of thermal resistances due to the following: • the thermal resistance due to heat transfer from the process fluid to cooler tube • the thermal resistance due to conductive heat transfer through the pipe walls • and the thermal resistance due to heat transfer from the cooler tubes to the seawater Creating heat transfer coefficients which for convection is the limiting factor com- 15 3. Theory pared to the conduction. The inner heat transfer coefficient varies with fluid com- position and velocity. For dense process fluid, the outer heat transfer coefficient is generally significantly smaller, which means it would govern the overall heat trans- fer [13]. While the inner and outer thermal resistances may be comparable for less dense fluids. The sum of the total thermal resistances between two fluids determines the value U or the overall heat transfer coefficient (OHTC). In order to relate this coefficient to heat transfer of an heat exchanger, the temperature difference must account for both the inner and outer heat transfer mediums, which is why the Log Mean Temperature Difference (LMTD) is used as defined below: ∆TLM = ∆T1 − ∆T2 ln ∆T1 ∆T2 (3.1) Where ∆T1 is the difference between the hot fluid inlet and the cold fluid outlet temperature, and ∆T2 is the difference between hot fluid outlet and cold fluid in- let. With this information the heat transfer can be calculated with the following equation: Q = UA∆TLM (3.2) Where Q is the heat transfer, A is the heat transfer area, and U is the OHTC. Natural convection or free convection as Incropera [14] refers to it as, occurs due to buoyancy forces within the fluid inducing natural currents to move through the fluid, compared to forced convection which is externally imposed. The buoyancy occurs with a fluid density gradient and a body force that is proportional to density, usually gravity. Since density of both liquids and gases are heavily dependent on temperature. This will be used for the seawater domain. Generally the Reynolds number is used on a fluid element to measure the ratio of the inertial to viscous forces, in a forced convection problem. However for free convection the Grashof number is used on a fluid to measure the ratio of the buoyancy forces to the viscous forces, which when no forced convection occurs will have the only effect (together with Prandtl) on the Nusselt number to characterise the convective heat transfer. 3.2 Heat Exchanger Design The simplest type of heat exchangers are a concentric tube with hot and cold fluid moving in the same or opposite direction. There are also heat exchangers where the fluid moves in cross flow over finned and unfinned tube banks. The most common industrial heat exchanger is the shell and tube where baffles are of importance to increase the convection coefficient of the shell-side fluid. Baffles induces turbulence and a cross flow velocity component relative to the tubes, additionally the tubes are supported by the baffles reducing flow-induced vibrations. [14] The shell and tube heat exchangers are active heat exchangers where a fluid is pushed through both the tubes and the shell from an external source. There are 16 3. Theory however passive heat exchangers where natural convection is utilised based on the temperature difference of the pipe fluid to the ambient surrounding fluid. For the subsea industry this is a useful tool as seawater has an immense cooling power. The active heat exchangers are currently the base case for offshore HVDC platforms as discussed in 2.2.1, where the seawater is pumped through the heat exchanger. This thesis however aims to introduce passive heat exchanger as a replacement for these as to reduce pump operations. Figure 3.1: Reference passive cooler Passive heat exchangers are investigated by Liu and Sakr [15] in order to find the parameters that greatly enhance the overall thermal performance. Generally for convection enhanced heat transfer is achieved with increasing effective surface area and residence time of the heat transfer fluid. There are several techniques to achieve this, passive heat exchangers are based on these principles. Techniques to increase heat transfer of passive coolers range from surface treatments and extensions, swirl flow devices and addatives to the fluids. However the conclusion of this investigation was that for turbulent flow passive techniques such as ribs, conical nozzle and ring are generally more efficient. 3.2.1 Pipe Arrangements The article by Khan, Zou and Yu [16] explored the characteristics of air-side heat transfer and pressure drop of staggered twisted tube bundles in cross-flow. Where the adjusted parameters can be referred to as transverse tube pitch ST , longitudinal tube pitch SL, diagonal tube pitch SD and longitudinal tube rows A1, which are clarified in figure 3.2. 17 3. Theory Figure 3.2: Example of tube arrangement parameters It was found when increasing the transverse tube pitch (from the base case being 2*Diameter) the Nusselt number and pressure drop decreases. The authors verified this finding with three separate studies which suggested that the turbulence flow in- creased with the increased pathway between the tubes, which gave lower maximum velocity, thus smaller transverse tube pitch is more appropriate with consideration of higher flow resistance. The effect of increase in longitudinal tube pitch increases the Nusselt number and pressure drop which coincided with another study found. As the longitudinal and diagonal tube pitches increases, the air distribution flows per- pendicularly to the twisted oval tubes, thus more proportional movement is allowed near the wall and mixing close to the tubes are enhanced. With more tube rows a notable increase was observed in both Nusselt number and pressure drop. Due to the increased surface area the performance of heat exchangers are dependent on the number of tube rows, which has been validated by two other studies [16]. 3.2.2 Chimney The work conducted by Kumar [2] investigated the physics of natural convection of air over a finned tube kept in a small chimney utilizing transient 3D numerical sim- ulations. The simulation model assumed the Boussinesq approximation was valid, which generally is the case for lower temperature cases. The geometry of the model can be seen in figure 3.3. 18 3. Theory Figure 3.3: Example of the chimney set-up by Kumar [2] The investigation of the chimney was in two parts, the first showed the effects of the chimney height which as the height increased resulted in greater heat transfer coefficient. The second part was how fin spacing affected the chimney effect and showed the smaller spacing between the tubes the chimney effect decreased. The chimney creates two major effects the first increases the driving force, and the second is a decrease in the air outlet temperature, with an increase in chimney height. The paper also investigated the transient behaviour of natural convection in the chimney, where the heat transfer is high initially, and then drops before reaching steady state. Which indicates that the simulations for natural convection should be more conservative. 3.3 Physics Models Computational fluid dynamics (CFD) is used to describe continuity, momentum, energy and species transfer between fluids. Traditional modelling base heavily on empericial and semi-empirical models, which often are not reliable for new process conditions. Therefore new design equations and parameters in existing models must be determined in order to accurately describe the changing equipment or process conditions outside the validated experimental database. Therefore CFD is a useful 19 3. Theory tool in describing continuity, momentum and energy transfer in a steady state sim- ulation. However in order to solve turbulence in steady state, transport equations are required. Single-phase laminar flow can be simulated accurately as well as for turbulent flows the simulations can be reliable. However for most cases and as is the case for this thesis complex turbulent flow are currently difficult to predict. Due to the proper- ties of turbulence the Navier-Stokes equations are used to describe the flow however even with super computers it is difficult to solve these complex engineering prob- lems. Especially with turbulence fluctuations where DNS and LES modelling would not handle the large amount of data, therefore most simulations use the Reynolds averaged Navier-Stokes (RANS) methods [8]. The paper by Paul [17] deep dives into different CFD models to accurately sim- ulate tube bundles. It was observed that the transverse turbulent intensity was significantly higher than the stream wise turbulent intensity. The overall perfor- mance of the models observed the k-based two equation models appeared closer to the experimental data than the particular second moment closure model (LRR-IP), especially the k-ε model were in agreement with the measured values. For the mesh generation k-ε and LRR-IP models showed less grid sensitivity than k-ω and SST models. Based on the discussions presented by the article and a previous course [8] it is known that for inside pipes where the walls are of interest the k-ω turbulence model is preferred, while surrounding the pipes the k-ε models are used. Decid- ing which k-based transport equations to use for each region was mainly based on internal knowledge. 3.3.1 EB k-ε The elliptic blending turbulence model solves the turbulent kinetic energy, k, and dissipation rate, ε, transport equations, the normalised (reduced) wall-normal stress component φ, and the elliptic blending factor α in order to determine the turbulent eddy viscosity. The elliptic relaxation was proposed by Durbin for Reynolds stress models [18]. The model was simplified for industry applications by Manceau and Hanjalic [19]. The transport equations: ∂ ∂t (ρk) + ∇(ρkv̄) = ∇ · [(µ 2 + µt σk )∇k] + Pk − ρ(ε − ε0) + Sk (3.3) ∂ ∂t (ρε) + ∇(ρεv̄) = ∇ · [(µ 2 + µt σε )∇ε] + 1 Te Cε1Pε − C∗ ε2ρ( ε Te − ε0 T0 ) + Sε (3.4) ∂ ∂t (ρφ) + ∇(ρφv̄) = ∇ · [(µ 2 + µt σφ )∇φ] + ρ ε0φt k0 + Pφ + Sφ (3.5) ∇ · (L2∇α) = α − 1 (3.6) 20 3. Theory 3.3.2 SST k-ω The k-ω turbulence model solves the transport equations for the turbulent kinetic energy, k, and the specific dissipation rate, ω which is the dissipation rate per unit turbulent kinetic energy ω ∝ ε k to determine the turbulent eddy viscosity. ∂ ∂t (ρk) + ∇ · (ρkv̄) = ∇ · [(µ + σkµt)∇ω] + Pk − ρβ∗fβ∗(ωk − ω0k0) + Sk (3.7) ∂ ∂t (ρω) + ∇ · (ρωv̄) = ∇ · [(µ + σωµt)∇k] + Pω − ρβ∗fβ∗(ω2 − ω2 0) + Sω (3.8) The SST variant model addresses a problem of sensitivity to free-stream/inlet con- ditions in the standard model remedied by Menter [20]. Where the following trans- port equations have an additional non-conservative cross-diffusion term containing the dot product ∇k · ∇ω in the ω production term. Pω = Gω + 2ρ(1 − F1)σω2 1 ω ∇k · ∇ω (3.9) 3.4 Boundary Conditions Boundary conditions are used to define the fluid domains, and the inlet, outlet and wall conditions are equally important for the simulation results as the differential equations. The usual condition for non-inlet/outlet boundaries is the wall bound- ary, which uses the ’no-slip condition’ stating the relative velocity between the wall and fluid is zero[8]. For heat transfer applications the wall can be defined according to heat transfer type, temperatures or heat flux, in this thesis the conjugate heat transfer condition is used as well as fixed temperature. Another important consideration regarding the Navier-Stokes equations is the initial guess. Since the equations are non-linear, the better initial conditions the better the solution will converge, demanding less computational time. If the start guess is out of order and the problem has several solutions the solution will diverge, which usually requires transient simulations to be solved. Stagnation inlet is used as an inflow condition used for both incompressible and compressible flows. It refers to the condition in an upstream imaginary plenum where the flow is completely at rest. For incompressible flows Bernoullis equation is used to relate total pressure, static pressure and velocity magnitude. Such as the boundary pressure given by: Ps = Pt,spec − ρ 2 |v|2 (3.10) The velocity magnitude is extrapolated from the interior of the domain: |v| = |v|ext (3.11) 21 3. Theory The velocity vector is given by: v = |v| · θspec (3.12) For a non-isothermal simulation the static temperature is calculated based on: Ts = Tt,spec − |v|2 2Cp (3.13) Where the density at the boundary face is then updated according to the static temperature: ρ = ρ(Ts) (3.14) The total and static enthalpies are given by: Hs = Hs(Ps, Ts) (3.15) Ht = Hs + |v|2 2 (3.16) The mass flow inlet, is for applying a specified mass flow at a inflow or outflow. The specified total mass flow rate is distributed over all faces of the boundary and calculates a uniform mass flow rate on each face. The pressure outlet imposes the working pressure, the boundary pressure can be considered the static pressure of the environment which the fluid enters. 3.5 Important Considerations When using Star CCM+ to simulate CFD problems many important considerations have to be taken into account. Which includes that the turbulence models use the All y+ wall treatment physics to describe the turbulent behaviour at walls. Other considerations are listed in the chapters below. 3.5.1 Solvers In order for the program to solve the transport equations, it must be selected if the flow equations should be solved for each component in a segregated, or cou- pled manner. Segregated meaning the flow equations are solved for each component pressure then each component velocity and are linked together the momentum and continuity equation with a predictor correction factor. While the coupled solver solves the conservation equations for mass and momentum together using a pseudo- time marching approach. The coupled solver require more memory however is more applicable for compressible flows and high Rayleigh number for natural convection. The segregated solver is actually designed for constant density flows, it can still be 22 3. Theory applied to compressible or low Rayleigh number natural convection flows. [21] Since the thesis will model a large 3D problem with a high computationally heavy load, the segregated solver is more applicable. Even though the coupled solver might converge in less iterations it requires a lot more fine tuning to achieve stability. The industry (internal) experience for this type of product (and temperature differences we have) have shown that the segregated solver converges faster but tend to under predict the thermal performance of the cooler. This means that the results will be on the conservative side and an increased performance is expected in reality. 3.5.2 Natural Convection Natural convection is the driving gravitational forces on a fluid due to density differ- ences in the domain. The density of the fluid is most often affected by temperature changes due to heat transfer. For this thesis this will be the driving force for moving seawater through the cooler. In order to accurately describe this phenomenon stag- nation inlet is applied, as well as some common practices highlighted by Siemens Star CCM+ team, which also has been verified in industry applications [22] • Ensure y+=1 at the heat transfer regions of the seawater domain. • For liquid flow a temperature dependent density should be used where poly- nomial density in function of temperature is the most convenient way, thus polynomial in T is utilised. • Selecting the solver is important, as coupled solver would be more robust for natural convection even if segregated solver is more conservative. The coupled solver is more difficult to use however recommended for high Rayleigh numbers. • More common practices are mentioned however based on this thesis these three are the most applicable for our model 3.5.3 Mesh Theory Creating a mesh in CFD is a complex procedure, but usually included in a commer- cial software packets for mesh generation. For 3D unstructured generation builds the mesh from different elements (tetrahedra, hexahedra, pyramids, prisms and do- cecahedra). An important consideration for the mesh generation is the walls which is why most applications require a surface remesher to be established before the volume mesh can start. In order to make sure the mesh is accurately representing the domain Andersson [8] suggests that using different grid spacing in the critical re- gions of the grid is a good way to avoid divergence of the results. Especially for flows with boundary layers which requires a dense mesh close to the wall, while regions far away from the mesh can be coarser. Andersson suggests these decisions will be based on intuition and previous knowledge, which for this thesis has been a crucial part in the decision making of the mesh generation. As the company has many years of experience with testing different settings and discovering what is reliable and not. 23 3. Theory 24 4 Methodology This chapter presents the general methods used for the thesis and the methodology regarding the market screening, cooling system analysis and how CFD was used. Design basis values for the CFD is also presented in this chapter. 4.1 Thesis Structure Since one of the initial aims of the thesis was to have dual perspective, combining both CFD and commercial aspects for the subsea cooler that where to be devel- oped, the philosophy behind the methodology was to initially have a broad focus that would narrow down to focus on solely the cooler as the thesis was progressing. The broader focus of the thesis consisted of screening competitive technologies that would pose a threat against the evaluated subsea cooling technology. The result of the screening resulted in a cost comparison between HVDC and HVAC that was used to screen the market for suitable HVDC platforms and establish the demand and cooling capacity. Following the market screening, a CFD analysis was conducted, investigating the cooler performance of the subsea cooler, by looking at the effect of adding baffles and how well the chimney performed. The conventional cooling system was also compared to the closed loop cooling system, utilising subsea cooling. Where the focus was on overall power consumption of the two systems and environmental im- pact. The remaining focus of thesis was aimed towards the subsea cooler based on results provided by CFD, where the original design was evaluated in terms of manufacturing possibilities. The entire cooler was split up into its sub-assemblies where every sub-assembly was treated independently in order for the analysis to be as comprehensive and structured as possible. In other words, the piping bundle was treated as one independent part as well as the shell and chimney encompassing the cooler bundle. 4.2 Literature Review To develop a deeper understanding of the topic of HVDC platforms, CFD analysis and subsea coolers a literature review was conducted. The literature review served as a basis for the thesis by integrating and comparing already established informa- tion around the topic with the results provided by the thesis. For a literature review to become trustworthy and possess a high level of credibility different approaches 25 4. Methodology can be taken [23]. For the thesis a semi-systematic literature review was selected as the most suitable approach, due to the somewhat lack of literature covering es- pecially subsea cooling for HVDC platforms. Subsea cooling for HVDC platforms is a new concept, hence the amount of published articles being limited. The semi- systematic literature is more suitable when the research questions are narrower and there is not enough literature available to conduct a systematic literature review [23]. The semi-systematic literature review was initiated by keyword searches as "Off- shore HVDC platforms", "Natural convection", "subsea cooling", etc., across Google Scholar and Chalmers Library as the main database. The first step was to screen titles and abstracts of selected articles to determine if they were relevant for the re- search. Secondly, if the articles were found relevant, they were further investigated and used in the thesis. In many cases the references in the selected articles were also used to dive deeper into the topic and gain even better accuracy for the research. 4.3 Data Collection and Data Processing The data used throughout the thesis consisted of both quantitative and qualitative data. The data typically origin from either published articles that were found rele- vant or from industry experts within the collaboration company through interviews. To gain information and data from within the company was crucial to provide re- liable results, since the topic in many aspects were in an early stage and therefore more difficult to find in published articles. This methodology strategy of using both qualitative and quantitative data was used to leverage the strengths of both types of data to address the research objectives to the largest extent. Since the aim of the thesis was to provide hands on numerical results with the use of CFD and costing estimates the mix of the two data types was required and is also a methodology that is widely used among several researchers [24]. Regarding data collection, the vast majority of the thesis relied on secondary data gathered from published sources, such as academic literature, industry reports from relevant companies within the offshore industry and government publications. Data used for the cost comparison between HVAC and HVDC was gathered from recent published bidding contracts. The reason secondary data was used as the main source of data was mostly due to the complexity and time consumption of gathering primary data. Since the evalu- ated cooler is intended for offshore subsea conditions it would require traveling to these locations to gather primary data for the design basis, which was outside of the limitations of the thesis. The same reasoning is applied to the market analysis and cost evaluation where the only data attainable is provided from other companies. However, the data was cross-checked between several sources and in many cases the average of numerical data was used to maintain credibility of the results, which is in line with what academic papers treating research methodology is suggesting [25]. The collected data was processed and capture in Microsoft Excel throughout the thesis. Meaning that all of the calculations for the market screening and cooling system analysis was done by this software. The same software was used to store the results of all the simulations, to use for comparison at a later stage in the thesis. 26 4. Methodology 4.4 Market Screening The research objective of the market analysis was to establish an understanding of the predicted market of size for HVDC platforms. The market screening was conducted as a progressive market screening, meaning that the scope of the market screening was continuously narrowed down to finally arrive at the HVDC platforms that were of interest for the thesis [26]. Since the HVDC market is correlated with the expansion of OWFs the initial data collection was aimed at gathering data over planned OWF. The research area was later narrowed down to focus solely on OWFs that planned to use HVDC instead HVAC. The market analysis had a prospec- tive perspective meaning that it emphasized a focus on the future, anticipating the potential outcomes and trends forming an outlook on the market for HVDC plat- forms. The data collection regarding information of planned OWF wind farms was accessible mainly through a database named "TGS 4C Offshore" that compiled data over OWFs that are operational, planned and under construction. TGS 4C Off- shore worked like an interactive map highlighting areas where different OWFs were positioned or planned to be positioned. The status of the OWF was indicated by different colours. Where green meant fully operational, orange under construction and light blue and purple that the consent for building an OWF had been autho- rised. The areas marked by a darker blue colour was specified as development, which meant that development was planned for post 2030. See Figure 4.1 for a snapshot of the map. When using the database, information about the various OWF would show up once an area was zoomed in and selected. Figure 4.1: Snapshot of TGS 4C offshore interactive map 27 4. Methodology Regarding the planned OWFs that were of most interest for the screening it was not specified which transmitting technology the wind farm was planning to use. To provide an estimate of which OWF that may be suitable for HVDC, research between HVAC and HVDC was conducted, with the aim of finding a break-even distance for when HVDC was more cost competitive than HVAC. The break-even distance established under Section 5.1.4 was then used a filter to screen out the OWFs that was closer to shore than the break-even distance and therefore assumed to be opting for HVAC rather than HVDC. 4.4.1 Cost Analysis Comparing HVAC and HVDC In this section of the methodology chapter, a comprehensive market screening was undertaken to examine current HVDC offshore platforms and forecast future projects, to establish a market outlook for the technology. Specifically, the investigation fo- cused on the North Sea region. The aim was to gather insights into existing HVDC platforms in the North Sea and understand why HVDC was preferred over HVAC. This information served as a basis for estimating the future demand for HVDC plat- forms. During the literature review, it was observed that HVDC’s primary advantage over HVAC is its ability to maintain transmitting voltage over very long distances, which is especially important for offshore wind. To determine which OWFs would benefit economically from HVDC rather than HVAC, a break-even distance was established. Previous studies suggested various distances at which HVDC was more cost-effective, but they focused on different applications and made assumptions that was not up- dated according to today’s climate. For instance, the electricity prices and discount rate used was no longer relevant. Therefore, a new analysis specifically targeting HVDC platforms for OWFs and comparing them with two types of HVAC transmit- ting technologies from a cost perspective, was conducted. The break-even distance was the result of collecting cost data and technical data from previously built plat- forms in the North Sea. The cost data for subsea cabling was collected through various published bidding contracts. Data on the costs of other relevant infrastruc- ture necessary for the analysis of both technologies was gathered from academic papers that had conducted similar studies in the same research area. Regarding electricity prices, the average electricity price between 2021 and 2023 for countries with coast to the north sea, excluding Norway was used. Norway was excluded since the majority of the energy mix consisted of hydro power and the price of electricity generated by wind power not being as relevant as other countries bordering to the North Sea. Initially the costs was split up between investment costs (CAPEX) and operating costs (OPEX). Where the investment costs covered cost associated with building the infrastructure required for both of transmitting technologies evaluated. The operat- ing cost consisted of transmitting losses, discounted over the life time of the OWF. Equation 4.1 was used to establish the present value of the future yearly OPEX to get a justified cost, considering the fact that these cost will occur in the future and 28 4. Methodology has to be put into the perspective of alternative investments. PV = 30∑ n=1 CFn · 1 − (1 + r)−n r (4.1) The sum of the two cost components made up the total cost and the final break- even distance. Microsoft Excel was utilized for data processing, calculations and visualization. The break-even distance served as a criterion to filter OWFs in the TGS 4C Offshore database, identifying those most likely to adopt HVDC technology. A list of HVDC platforms from its associated OWF meeting the distance criteria was compiled as a bar chart as the outcome of the market analysis. The outcome was further analysed by computing the average converting capacity for the identified projects and converting it into cooling demand. 4.5 Cooling System Analysis The cooling system analysis was focusing on both the environmental impact of the system and the actual power consumption between the systems. To determine the power and cost saving of choosing a closed loop rather than an open loop that pumps sea water to the platform the different eliminated components for the closed loop was further investigated. Firstly, the eliminated sea water lift pumps was analyzed where calculations was made in order to find out how much energy that was required to pump the sea water up to the platform where the heat exchanger was positioned. The circulation pumps was also investigated, together with the electrochlorination unit to determine the total energy consumption of the cooling system which resulted in how much the electricity that would have to be scavenged from the wind farm in order to keep the cooling system functional. Equation 4.2 was used to determine the work required to lift the sea water or cooling medium in both of the cases. P = ṁ · g · h · η (4.2) To calculate the energy consumption regarding the electrochlorination unit, equa- tion 4.3 was used. TE represent the total energy to produce sodium hypochlorite at a specific mass flow, expressed in watts. Eprod is the energy required to produce one kilogram of sodium hypochlorite through electrochlorination for using seawater and is expressed in kWh/kg. The energy consumption was based industrial equip- ment’s commonly used in cooling systems utilizing seawater as cooling medium, were biofouling mainly occurs. TE = ṁ · Eprod (4.3) To represent the cost associated with the required work, the same price of electric- ity established under the market analysis was used, see Table 5.2. The same values were used to maintain consistency with the previous results provided by the thesis, thereby ensuring uniformity and comparability throughout the analysis. The yearly energy consumption was based on the assumption that the system was constantly operational, 24 hours a day and 365 days a year. In reality the system would prob- ably have some kind of downtime, due to maintenance or failure. But since the aim 29 4. Methodology of the thesis was to compare the two systems and an added scope of investigating the up-time of the two systems was considered to be outside of the limitations. 4.6 Design Basis The design basis for this project was based on research and discussions with experts. It was divided into material, cooling water loop, and seawater. Since the aim of the thesis was to design a cooler applicable for as many places as possible, the design and testing simulations were done in worst-case conditions. Even though anchored wind farms are usually around a maximum depth of 100m, the cooler should not be excluded to these designs as floating wind farms should also be considered, which have a maximum depth of 300m. Another constraint regarding the design of the cooler was a height restriction, which was considered for the chimney design. 4.6.1 Material Since the material of the piping tubes are requested to move away from Cu-Ni, stain- less steels are appropriate replacements, even though their heat transfer properties are known to be worse, but better corrosion resistance. Stainless Steel has been used as base case for material and was used throughout the whole thesis, due to its high corrosion protection. Table 4.1: Pipe material specification Parameter Value Unit Density 8000 kg/m3 Specific heat capacity 502 J/kg-K Thermal Conductivity 15.9 W/m-K The pipe dimensions were designed after a 300m ocean depth, to handle all types of wind farm applications. 4.6.2 Cooling water Loop The cooling water loop consisted of water with 10% MEG. The properties of the medium was provided by internal company knowledge, and the cooling medium was not further evaluated in this thesis. The mass flow of the medium was assumed based on internal knowledge, since the only constraint was the ability to fill each pipe stack without back pressure. The inlet and outlet constraints for the converter platforms were discussed with experts and found to differ between products and companies, however a range was provided and used as guideline. A temperature difference of 10-20K should be assumed between inlet and outlet. For the outlet (to the cooler) the range was 25-30°C, while for the inlet the range was 35-45°C. With 30 4. Methodology these parameters the specifications were established as in table 4.2, with the aim to create a worst-case scenario. Table 4.2: Cooling fluid specification Parameter Value Unit Inlet Temperature 35 °C Desired Outlet Temperature 25 °C Mass flow 2.2 kg/s-pipe 4.6.3 Seawater properties Due to the subsea cooler being designed for worst case scenario all parameters were the max or min of possible outcomes, which means no seawater velocity, a high ambient temperature and salinity. The properties of seawater was provided by internal knowledge, and the specification can be seen in table 4.3. Table 4.3: Seawater specification Parameter Value Unit Ambient Seawater Temperature 20 °C Current velocity 0 m/s Salinity 35 % 4.7 Building the Model The dimensions of the coolers were provided by company in a Solidworks assembly, with multiple solidwork parts, some were considered unnecessary for the CFD anal- ysis (see chapter 2.3) and were removed or simplified as to reduce computational time. The simplified parts included the large bend to the outlet header and the out- let header, while the walls and inlet header were removed in the seawater domain. In order to build the model, trial and error was utilised, where the thesis attempted to build the model from scratch at first, however due to the amount of parame- ters affecting the inlet and outlets of the pipe geometry, it was decided to use the provided pipe parts. Since the pipes are staggered and the inlet and outlet of the pipe are placed differently. The two different pipe stacks were extracted as parts in Solidworks, and assembled without any other part, to then be patterned to the amount of pipe stacks required for the pipe bundle for each simulation. 31 4. Methodology Figure 4.2: Subsea cooler domain setup The Solidworks assembly with the pipe bundle were then imported into SpaceClaim 2023 R2, which was company standard to import to 3D CAD in Star CCM+. A scene specifying each domain can be seen in figure 4.2. The volume inside the pipes were added through volume extract, and the seawater domain was sketched out and using the ’pull’ tool created around the pipe bundle. 4.8 Meshing the model The mesh controls were based on a previous study on the cooler performed by company. The automated mesh utilised surface remesher, polyhedral mesher and prism layer mesher. To specify the multitude of domains with specific requirements the default controls were lowered slightly however all surfaces were specified to a custom control listed in table 4.4. These settings were used for the model, however some parts were more critical than others which was why refinement boxes were used as well. The placements of these can be seen in figure 4.2 where the custom size was applied slightly smaller than the seawater. Another refinement box was created directly in Star CCM+ to simulate the plume from the outlet of the cooler. There were some issues with the mesh from the previous study due to the pre- carious design of the pipes, the mesh inside and surrounding the pipes were piercing each other therefore the surface size had to be reduced. The mesh settings speci- fied above was finer than previous study, however to check the sensitivity to grid 32 4. Methodology Table 4.4: Automated mesh custom controls Surface Custom Control Value 1. Cooling Fluid 2. Seawater Pipes Surface Target Surface Size Base Size Number of Prism Layers 4 Prism Layer Total Thickness 2.8416mm Solid Pipes Target Surface Size Base Size Prism Layer Disabled Seawater: Cooler Walls Inlet Header Target Surface Size +50% Number of Prism Layers 2 Prism Layer Total Thickness 10mm Seawater Target Surface Size 0.4m Number of Prism Layers 2 Prism Layer Total Thickness 0.2m refinement a mesh study was conducted. Additional prism layers were applied and compared to the original mesh, to see the effects of smaller y+ value inside the pipes would improve or hinder the heat transfer to the seawater. A further refinement of the target surface size was also performed to the effects of refinement outside the pipes. Since the surface size governs how well the circular pipe is built with tetra- hedral shapes, if the tetrahedral are smaller more area would be added to create the mesh more accurately describe the pipe surface. Therefore the heat transfer area was expected to increase with smaller surface size, which as introduced in chapter 3.5.3, would benefit the heat transfer to the seawater. 4.9 Simulation Setup The simulation setup varies depending on what was simulated and was specified in the sections below, however the physics of each medium remains the same. Unless specified no parameters were changed from the default settings of the physics mod- els applied. All physics are three dimensional, steady state and using a segregated solver. Cooling water medium was specified as a Fluid continuum and the SST K-ω model was set as turbulence model based on recommendations from company. The SST k- omega works well with adverse pressure gradients and was unsuited for the seawater region as it tends to over predict turbulence in regions with large normal strain. The discretization scheme used was specified by the turbulence models default settings. The liquid properties of the medium were modelled using polynomial functions de- pendent on temperature according to the design basis. The region for the cooling water has mass flow inlet and pressure outlet as boundary conditions, with the no- slip wall condition however contact interfaced with the solid pipe material with a conjugate heat transfer condition. The pipe material was Stainless Steel for which a solid physics continuum was used, 33 4. Methodology with constant material properties specified in the design basis. The solid pipe re- gion was connected towards the cooling fluid and seawater regions using conformal interfaces that model the conjugate heat transfer across the boundaries. For the seawater surrounding the pipes it was decided to use EB k-ε turbulence model based on company experience when modeling natural convection for this type of application. As the flow around the pipes were the important areas as well as the turbulence modelling being more robust. As the cooling fluid material prop- erties, polynomial temperature based functions were used to simulate the natural convection, as well as the all y+ wall treatment being of importance. The regions for the seawater utilises stagnation inlet and no slip wall for the seabed. It also had an interface towards the solid pipes. 4.9.1 Pipe arrangements In order to get started with simulations it was decided to test a simplified model with only 4 pipes with varying parameters. Due to the computational demand of the full size cooler it was decided to test these configurations on a reduced model. The SolidWorks 3D-model that formed the basis of the design did not contain any fluid domains. As such, the external seawater domain as well as internal cooling water domain was designed in Spaceclaim. This model was then imported into Star ccm+ where the computational model was built. This model included 4 pipe stacks and a cylindrical seawater domain with the same radius as the cooler walls. The volume extruder mesh function was used to create larger domains above and below the imported pipes. The domain was changed two times as issues with convergence occurred, instead of having a cylindrical seawater domain, a cone was placed above the pipes as to create a smaller outlet surface. This issue was due to the pressure outlet boundary as the outlet surface was to large to find a steady state solution and large areas of back flow occurred which was solved by reducing the outlet size. The two domains can be seen in figure 4.3. Figure 4.3: Pipe arrangement seawater set-up 34 4. Methodology Two designs were thought of to test the first simulations. The first adding more spirals towards the inlet header to increase heat transfer area and prevent channeling. The second increasing and decreasing the longitudinal pitch, as suggested in the literature study. The design tests can be seen in figure 4.4. Figure 4.4: Pipe arrangement designs 4.9.2 Full 3D model In order to verify results of the simplified models the thesis conducted three large scale simulations, one with only walls, one with the chimney, and one with chimney and baffles. The large simulation model utilised the pipes built in Solidworks and created an assembly with 20 staggered pipes. The assembly was then imported to SpaceClaim where walls, inlet header, seawater domain, and cooling water in pipes were added. The seawater domain was created as a cylinder with a diameter of 4 times the cooler’s, and the same height with half a sphere above with a radius of 2 times the cooler. The boundary conditions in Star CCM+ was the same as described in chapter 4.9, the cooling water had mass flow inlet and pressure outlet. Figure 4.5 displays all boundary conditions applied to the large domain, where cooling fluid inlet and outlet refers to all pipes in the bundle, as the boundary condition was the same for each inlet and outlet. The seawater has stagnation inlet on the sphere and cylinder walls, the bottom circle was considered the seabed which has the wall no-slip condition. The solid pipe has only the wall condition with conjugate heat transfer to both fluid sides. 35 4. Methodology Figure 4.5: Boundary conditions on the cooler The subsea cooler 3D simulations evaluated the following three cases to benchmark trends: • Wall: Only containing the walls • Chimney: Containing the walls and chimney • Baffles: Containing the walls, chimney and baffles around the inlet header As stated in chapter 4.7 when the walls and chimney were included they were sub- tracted from the seawater domain as no information except the seawater flow around the geometry was necessary for the cooler performance. 4.9.3 Model Simplifications The full 3D model represents a significant computational demand due to the size of the cooler, the computational mesh included above 120 million cells. In order to resolve the physics of the cooler and provide results with a sufficient level of accu- racy it was necessary to do model simplifications. As one of the research questions stated, the thesis attempted to investigate simplification methods in order to design the chimney, as this would only require the seawater domain. At first the chimney was attempted to be designed based on a 2D model of the domain, however after verification the design was invalidated, and another approach of taking a wedge of the 3D model was utilised. In order to do this the thesis only simulated the seawater domain, and assigned temperatures to the outer heat transfer surfaces. The temperatures were extracted from simulation "wall" according to fig- ure 4.6, where the top two and bottom two pipe stack surface average temperatures were reported from the seawater surface around the pipes. These were assumed to 36 4. Methodology represent the average temperature of each half spiral surface, as there were not much difference in between pipe stacks (y-direction). Figure 4.6: Temperature profiles from simulation wall These temperatures weren’t meant to be accurate they were meant to provide a fixed temperature value, when designing the chimney, in order to see trends to assess the heat transfer to the rest of the seawater domain. To design the best chimney. 4.9.3.1 2D model The chimney design, was decided to be designed based on a 2D model for only the seawater domain. Since the cooler pipes are spirals the tubes and cooling medium cannot accurately be represented, the temperature profiles were applied to the sea- water surfaces. The challenge arose when applying the temperature data on each seawater hole surface, where firstly all surfaces were named from top 1-19 and from the inlet header 0-20 as a matrix, see figure 4.7 creating 390 surfaces to be assigned a temperature. This was done in the seawater region through 4 boundaries for each pipe data set, with surface sub-groupings filtering for each pipe surface into Pipe.x.y, which was created with a java script to avoid having to filter and name each sub- group 100 times. The thermal specification was set to fixed temperature and a Java script was utilised which looped the temperatures and assigned them to the correct surface based on their names. 37 4. Methodology Figure 4.7: 2D model surface naming matrix It is important to note that the 2D model in star ccm, is not purely 2D it accounts for one cell depth. To model 2D in star ccm+ a two dimensional surface representation was imported which is then extruded 1 mesh cell in the z direction, in order to create a 2.5D. To adjust and design the chimney with ease it was decided for it to be designed in 3D CAD in Star CCM+ with a sketch and cut extrude. In order to verify this simplification a copy of the simulation "Wall", simulation "Chimney" and simulation "Baffles", was created and the trends were compared. 4.9.3.2 Wedge model The wedge model was the second attempt of simplifying the designing of a large domain simulation. For this case the original geometry for the full simulation was cut at first a 11.25 degree angle, and then a 22.5 degree angle. A similar simulation approach as was used for 2D was used for the wedge model, meaning only the seawater domain was considered and a temperature profile was specified on the pipe surfaces. To simulate the heat transfer from the pipes, the thesis decided against renaming all 390 surfaces as was done in the 2D model, instead the top ten pipe rows were renamed pipes1 and the bottom ten pipe rows were named pipes2. Two temperature profiles were imported to Star CCM+ as tables, table T1 and table T2, which can be seen in figure 4.8. These temperatures were assigned to each pipe surface based on its radius according to a field function. Even though this function was not exact since it applied different temperatures around the pipe the differences were considered negligible since the alternative to do as the 2D model would take too much time. 38 4. Methodology Figure 4.8: Temperature distribution on pipe surfaces The wedge model were verified in the same matter as the 2D model with three simulations according to simulation "Wall", simulation "Chimney" and simulation "Baffles". 4.10 Final Model Design In order to test cooler the following parameters were changed and evaluated. 4.10.1 Chimney Design When designing the chimney the following parameters were tested as can be seen in figure 4.9. Where the chimney visible on the left hand side was the original design and Sim refers to the simulation number that was run, which changes a certain constraint. 39 4. Methodology Figure 4.9: Wedge tests designs These tests were designed based on the streamlines scalar presented in chapter 5.5.2, where in the first tangent arc of the chimney flow gathered and swirling regions occurred, thus these tests are designed to attempt to prevent these. • Sim 1: Decreasing the radius of both tangent arcs • Sim 2: Increasing the chimney height and decreasing the chimney outlet • Sim 3: Increasing the chimney height (with smaller radius of the bottom arc) • Sim 4: Increasing the chimney height, decreasing the bottom arc, and increas- ing the top arc • Sim 5: Decreasing the bottom arc, and increasing the top arc 4.10.2 Design Improvements Based on the outcome for the 2D and wedge simulations full 3D model simulations were performed to test the robustness of the cooler. Based on those full 3D simula- tion results it was decided to attempt to remove some pipe spirals. Based on the result presented in chapter 5.5, concerns regarding no change in ef- fect with chimney or with only walls were raised. Also the outlet temperature in the pipes did not differ much from each other from the last three spirals, the the- sis wanted to test if removing five outer spirals would increase the chimney effect. Another parameter to change was the inlet temperature of the cooling fluid as this could increase the chimney effect but increase the outlet temperature out of the desired range. Therefore the following models were tested and compared using the 40 4. Methodology same simulation set up as chapter 4.9.2: • "No Walls" Simulation with only tube bundle • "Baffles" Simulation with walls, chimney and baffles • "BafflesT45" Simulation with walls, chimney and baffles with inlet tempera- ture of cooling fluid increased to max of 45 °C • "HTA70"Simulation with walls, chimney and baffles with 5 spirals removed from the outside • "HTA70T45" Simulation with walls, chimney and baffles with 5 spirals re- moved from the outside inlet temperature of cooling fluid increased to max of 45 °C 4.10.3 Post-processing The post processing of the models was based on reports for multiple scalar or vector functions pre-defined in the Star CCM+ database. The Overall heat transfer coefficient was calculated using LMTD however by conven- tion, for this type of heat exchanger, the inlet and outlet temperature was assumed to be the same. Giving the following definition of LMTD, ∆TLM = Tcw,in − Tcw,out ln Tcw,in−Tambient Tcw,out−Tambient (4.4) Where Tcw,in is the cooling water inlet temperature, Tambient is the seawater ambient temperature, and Tcw,out is the surface average outlet temperature of all the pipe outlets. Which was then used for the following equation OHTC = Q A∆TLM (4.5) Where the report for heat transfer between the sea and pipe is Q, the mesh area report of the OD pipe is the heat transfer area A. The result is U or OHTC, which in this report is presented as a normalised value. Another parameter that had a special set-up in the simulations were the measure of mass flow of seawater through the cooler. This was computed through a mass flow report, based on a plane section created from a threshold derived part with the cooler walls diameter. All the scenes extracted from the simulations to be compared were normalised. The simulation with the highest and lowest scalar function value was applied to all scenes to be compared. This was true for all scenes except those presented in Chap- ter 5.7.1, since the timing of the simulations caused all the scenes to be extracted before sim 5, the scenes are normalised but based on sim 4, even though sim 5 had higher/lower scalar values. 41 4. Methodology 4.11 Manufacturing Possibilities When assessing the product cost from a manufacturing perspective of the subsea cooler the input from the CFD analysis was used. To determined which sub- assemblies which was of most interest to investigate. Since the CFD analysis in- vestigated the piping bundle and different configurations of the pipe stacks and what effect the chimney had on the heat transfer these two sub-assemblies was fur- ther investigated from a manufacturing perspective. It was evident that the cooler design would allow for a lower total amount of welds and therefore literature re- searching welding and cold bending was used as empirical material to be assessed on the cooler of what cost implications could be expected when a large amount of welds was replaced with a continuous bend. The 3D model was examined in Solid- Works together with other subsea coolers with a rectangular design, similar to the one illustrated in 3.1. A similar methodology was followed for the manufacturing investigation regarding the chimney design. The chimney was also examined in SolidWorks and literature was reviewed in order to provide empirical evidence behind how the chimney could be manufactured. Results from the CFD was also used as a basis for the discussion whether implementation of a chimney was worth it or not. Input in form of internal knowledge regarding the manufacturing of both the chimney and the pipe bundle was also used. 42 5 Results & Discussion In this section the result from all parts of the thesis is presented and discussed. 5.1 Market Screening The following section provides the results of the technical market analysi