2017 

Hawkes, James Chaotic Methods for the Strong Scalability of CFD PhD Thesis University of Southampton, 2017. Abstract  Links  BibTeX  Tags: Chaotic Solvers, Exascale, Parallelization, RANS, Solvers @phdthesis{2017PhDJamesHawkes, title = {Chaotic Methods for the Strong Scalability of CFD}, author = {James Hawkes}, url = {http://www.refresco.org/download/2017phdjameshawkespdf/}, year = {2017}, date = {20170824}, school = {University of Southampton}, abstract = {Supercomputing power has been doubling approximately every 14 months for at least three decades, increasing the capabilities of scientic modelling at a similar rate. The first machines capable of one ExaFLOP (1018 foatingpoint operations per second) are expected by 2020. However, architectural changes required to reach `exascale' are significant, with energy efficiency constraints leading to a huge growth in parallelization. A new era of computing has arrived, dubbed the `manycore' era, in which the number of computing cores is increasing faster than CFD simulation sizes { prompting the research question for this thesis: `What limits the strong scalability of CFD and its ability to handle manycore architectures? What can be done to improve the CFD algorithms in this respect?' A number of scalability investigations have been performed from 1 through to 2048 cores, using a semiimplicit, finitevolume CFD code: ReFRESCO; and the University of Southampton supercomputer: Iridis4. The main bottleneck to strong scalability is shown to be the linear equationsystem solvers, occupying up to 95% of total walltime on 2048 cores { where the poor scalability arises from synchronous, global, interprocess communications. Experiments have been performed with alternative, stateoftheart linear solvers and preconditioners, without significant improvements, which motivates novel research into scalable linear solvers for CFD. The theory of `chaotic relaxation' has been used to create a completely asynchronous Jacobilike `chaotic solver', showing almost perfect scalability, and performance far greater than their synchronous counterparts. However, these solvers lack the absolute numerical power to compete with existing solvers, especially as the resolution of the simulations increases. Following this, chaotic relaxation theory has been used to create a novel `chaoticcycle' multigrid solver, combining aspects of the chaotic solver and classical multigrid methods. Both of the solvers have been verfiied and tested using canonical test cases and practical CFD simulations. On 2048 cores, the chaoticcycle multigrid solver performs up to 7:7x faster than a typical Krylov Subspace solver and 13:3x faster than classical Vcycle multigrid. With improvements to the implementation of coarsegrid communications and desynchronized residual computations, it is likely that the chaoticcycle multigrid method will continue scaling to many thousands of cores, thus removing the main bottleneck to the strongscalability of CFD. The novel chaotic solver and chaoticcycle multigrid methods have been implemented as an opensource library, Chaos. It is hoped that work on these scalable solvers can be continued and applied to other disciplines.}, keywords = {Chaotic Solvers, Exascale, Parallelization, RANS, Solvers}, pubstate = {published}, tppubtype = {phdthesis} } Supercomputing power has been doubling approximately every 14 months for at least three decades, increasing the capabilities of scientic modelling at a similar rate. The first machines capable of one ExaFLOP (1018 foatingpoint operations per second) are expected by 2020. However, architectural changes required to reach `exascale' are significant, with energy efficiency constraints leading to a huge growth in parallelization. A new era of computing has arrived, dubbed the `manycore' era, in which the number of computing cores is increasing faster than CFD simulation sizes { prompting the research question for this thesis: `What limits the strong scalability of CFD and its ability to handle manycore architectures? What can be done to improve the CFD algorithms in this respect?' A number of scalability investigations have been performed from 1 through to 2048 cores, using a semiimplicit, finitevolume CFD code: ReFRESCO; and the University of Southampton supercomputer: Iridis4. The main bottleneck to strong scalability is shown to be the linear equationsystem solvers, occupying up to 95% of total walltime on 2048 cores { where the poor scalability arises from synchronous, global, interprocess communications. Experiments have been performed with alternative, stateoftheart linear solvers and preconditioners, without significant improvements, which motivates novel research into scalable linear solvers for CFD. The theory of `chaotic relaxation' has been used to create a completely asynchronous Jacobilike `chaotic solver', showing almost perfect scalability, and performance far greater than their synchronous counterparts. However, these solvers lack the absolute numerical power to compete with existing solvers, especially as the resolution of the simulations increases. Following this, chaotic relaxation theory has been used to create a novel `chaoticcycle' multigrid solver, combining aspects of the chaotic solver and classical multigrid methods. Both of the solvers have been verfiied and tested using canonical test cases and practical CFD simulations. On 2048 cores, the chaoticcycle multigrid solver performs up to 7:7x faster than a typical Krylov Subspace solver and 13:3x faster than classical Vcycle multigrid. With improvements to the implementation of coarsegrid communications and desynchronized residual computations, it is likely that the chaoticcycle multigrid method will continue scaling to many thousands of cores, thus removing the main bottleneck to the strongscalability of CFD. The novel chaotic solver and chaoticcycle multigrid methods have been implemented as an opensource library, Chaos. It is hoped that work on these scalable solvers can be continued and applied to other disciplines.  
2016 

Gharraee, Behrad Numerical Simulation of Cavitation on a Tidal Turbine using ReFRESCO Masters Thesis Chalmers University of Technology, 2016. Abstract  Links  BibTeX  Tags: Cavitation, Current Turbines, KSKL, RANS, SST, URANS, Verification @mastersthesis{2016Msc_Thesis_Gharraee, title = {Numerical Simulation of Cavitation on a Tidal Turbine using ReFRESCO}, author = {Behrad Gharraee}, url = { http://www.refresco.org/download/2016msc_thesis_gharraeepdf/}, year = {2016}, date = {20160104}, address = {Gothenburg}, school = {Chalmers University of Technology}, abstract = {As renewable energies continue to grow their share in the global energy landscape, marine resources present an inexhaustible potential to provide the ever increasing human settlements energy demands. Tidal energy conversion technologies enjoy the benefits of the accurately predictable and highly reliable resources, while promising great power to weight ratio due to the relatively small size of the equipment compared with offshore wind for instance. There are various prototypes being tested today and some proposals are employing floating structures as the platform for the energy converters, the design of which is driven by the higher kinetic energy content of the streams close to the water surface. Such concepts increase the turbines susceptibility to cavitation. There has been very little explicit research performed on the cavitation behavior of tidal turbines and this thesis attempts to establish one such study to enable and promote future investigations. The specialized hydrodynamic RANS solver ReFRESCO is used with the builtin Sauer cavitation model. Structured grids have been employed. The effectiveness of an eddyviscosity modification method known as the Reboud correction is also subject of investigation for improving dynamic behavior of cavities. Two different turbulence models used are kOmega SST (SST2003) and kskL. A threebladed model scale Horizontal Axis Tidal Turbine (HATT) is numerically simulated in openwater conditions in an attempt to reproduce previous EFD results from the University of Southampton, thus validating the numerical procedures in use. The simulations are performed through three stages where initially a steady solution is obtained, then the simulation becomes transient and finally the cavitation model is switched on. The results are validated against experiments via nondimensionalized parameters for thrust and torque, which prove satisfactory. General flow shows good agreement with experimental observations and the cavity formation appears to be accurate regarding both its position and blade coverage. Interestingly a cavity is observed near the leading edge on the pressure side. The simulations fail to resolve the details near the closure line of the sheet cavity which is attributed to inadequate meshing resolution. Very little dynamic behavior of the cavity structure is observed specifically where a "horseshoe" cavity structure had been detected during EFD, which will be subject to future work.}, keywords = {Cavitation, Current Turbines, KSKL, RANS, SST, URANS, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } As renewable energies continue to grow their share in the global energy landscape, marine resources present an inexhaustible potential to provide the ever increasing human settlements energy demands. Tidal energy conversion technologies enjoy the benefits of the accurately predictable and highly reliable resources, while promising great power to weight ratio due to the relatively small size of the equipment compared with offshore wind for instance. There are various prototypes being tested today and some proposals are employing floating structures as the platform for the energy converters, the design of which is driven by the higher kinetic energy content of the streams close to the water surface. Such concepts increase the turbines susceptibility to cavitation. There has been very little explicit research performed on the cavitation behavior of tidal turbines and this thesis attempts to establish one such study to enable and promote future investigations. The specialized hydrodynamic RANS solver ReFRESCO is used with the builtin Sauer cavitation model. Structured grids have been employed. The effectiveness of an eddyviscosity modification method known as the Reboud correction is also subject of investigation for improving dynamic behavior of cavities. Two different turbulence models used are kOmega SST (SST2003) and kskL. A threebladed model scale Horizontal Axis Tidal Turbine (HATT) is numerically simulated in openwater conditions in an attempt to reproduce previous EFD results from the University of Southampton, thus validating the numerical procedures in use. The simulations are performed through three stages where initially a steady solution is obtained, then the simulation becomes transient and finally the cavitation model is switched on. The results are validated against experiments via nondimensionalized parameters for thrust and torque, which prove satisfactory. General flow shows good agreement with experimental observations and the cavity formation appears to be accurate regarding both its position and blade coverage. Interestingly a cavity is observed near the leading edge on the pressure side. The simulations fail to resolve the details near the closure line of the sheet cavity which is attributed to inadequate meshing resolution. Very little dynamic behavior of the cavity structure is observed specifically where a "horseshoe" cavity structure had been detected during EFD, which will be subject to future work.  
2015 

Rosmarino, Alessandra PREDICTION OF HYDRODYNAMIC PERFORMANCES OF SAILING YACHTS, COMPUTED USING CFD CODES Masters Thesis University of Study of Genova, Italy, 2015. Abstract  Links  BibTeX  Tags: Drag, Lift, Manoeuvring, RANS, SST, Validation, Yachts @mastersthesis{2015Msc_Thesis_ARosmarino, title = {PREDICTION OF HYDRODYNAMIC PERFORMANCES OF SAILING YACHTS, COMPUTED USING CFD CODES}, author = {Alessandra Rosmarino}, url = {http://www.refresco.org/?p=1397}, year = {2015}, date = {20150331}, school = {University of Study of Genova, Italy}, abstract = {The present work deals with the prediction of performance of sailing yachts as computed with inhouse CFD codes. The main part of this study is represented by the process made in order to run steady computations regarding two hulls, sailing with different heel and yaw angles. Since ReFRESCO does not compute the free surface and neither dynamic trim and sinkage, several tools such as RAPID for the free surface generation have been used. Finally the results show that is possible to predict the drag with an accuracy of 2%, and good results can also be obtained for the lift. Further studies are required with a higher refinement level especially close to the keel, and investigations on the influence of different parameters on the quality of the wave pattern obtained from RAPID are necessary. Furthermore a trade off study has to be done in order to assess if the higher accuracy expected from a total RANS approach will compensate the higher computing effort that will be required.}, keywords = {Drag, Lift, Manoeuvring, RANS, SST, Validation, Yachts}, pubstate = {published}, tppubtype = {mastersthesis} } The present work deals with the prediction of performance of sailing yachts as computed with inhouse CFD codes. The main part of this study is represented by the process made in order to run steady computations regarding two hulls, sailing with different heel and yaw angles. Since ReFRESCO does not compute the free surface and neither dynamic trim and sinkage, several tools such as RAPID for the free surface generation have been used. Finally the results show that is possible to predict the drag with an accuracy of 2%, and good results can also be obtained for the lift. Further studies are required with a higher refinement level especially close to the keel, and investigations on the influence of different parameters on the quality of the wave pattern obtained from RAPID are necessary. Furthermore a trade off study has to be done in order to assess if the higher accuracy expected from a total RANS approach will compensate the higher computing effort that will be required.  
2014 

Saraiva, Goncalo Solution of Flows Around Airfoils Using RANS with WallFunctions Masters Thesis IST, Lisbon, Portugal, 2014. Abstract  Links  BibTeX  Tags: Eppler, Foils, NACA 0012, RANS, SST, Validation, Verification, Wallfunctions @mastersthesis{2014Msc_Thesis_GoncaloSaraiva, title = {Solution of Flows Around Airfoils Using RANS with WallFunctions}, author = {Goncalo Saraiva}, url = { http://www.refresco.org/?wpdmpro=2014msc_thesis_goncalosaraivapdf}, year = {2014}, date = {20141024}, school = {IST, Lisbon, Portugal}, abstract = {The calculation of the friction forces is essential in hydrodynamic and offshore applications. However, the high gradients that exist in nearwall regions require the use of one of the following approaches: grids that are very fine near the wall to calculate the wall shearstress directly from its definition; or wallfunctions (WF) to calculate indirectly the wall shearstress and provide boundary conditions for the variables of the turbulence models. The objective of this thesis is to assess the validity of WF boundary conditions for the calculation of friction and pressure coefficients, as well as aerodynamic forces coefficients of a conventional and a laminar airfoil. The ReFRESCO solver was used to solve the RANS equations with the SST version of the eddyviscosity turbulence model. The main conclusions obtained were: WF can yield acceptable results if the Reynolds number is high enough to promote transition near the leading edge; if the laminar part of the flow is significant, the results are not realistic because WF lead to a fully turbulent flow; the results for the pressure and lift coefficient are always better than for friction and drag coefficients due to the direct connection of the wall shearstress with the last two; last but not least, the results of the WF approach are strongly dependent on the location of the first interior grid node, even at high Reynolds number.}, keywords = {Eppler, Foils, NACA 0012, RANS, SST, Validation, Verification, Wallfunctions}, pubstate = {published}, tppubtype = {mastersthesis} } The calculation of the friction forces is essential in hydrodynamic and offshore applications. However, the high gradients that exist in nearwall regions require the use of one of the following approaches: grids that are very fine near the wall to calculate the wall shearstress directly from its definition; or wallfunctions (WF) to calculate indirectly the wall shearstress and provide boundary conditions for the variables of the turbulence models. The objective of this thesis is to assess the validity of WF boundary conditions for the calculation of friction and pressure coefficients, as well as aerodynamic forces coefficients of a conventional and a laminar airfoil. The ReFRESCO solver was used to solve the RANS equations with the SST version of the eddyviscosity turbulence model. The main conclusions obtained were: WF can yield acceptable results if the Reynolds number is high enough to promote transition near the leading edge; if the laminar part of the flow is significant, the results are not realistic because WF lead to a fully turbulent flow; the results for the pressure and lift coefficient are always better than for friction and drag coefficients due to the direct connection of the wall shearstress with the last two; last but not least, the results of the WF approach are strongly dependent on the location of the first interior grid node, even at high Reynolds number.  
Make, Michel Predicting scale effects on floating offshore wind turbines Masters Thesis Technical University of Delft, the Netherlands, 2014. Abstract  Links  BibTeX  Tags: BEMT, Foils, MSWT, NREL 5MW, RANS, ScaleEffects, Scaling, SpalartAllmaras, SST, Transition, Turbines, URANS, XFOIL @mastersthesis{2014Msc_Thesis_MichelMake, title = {Predicting scale effects on floating offshore wind turbines}, author = {Michel Make}, url = { http://www.refresco.org/?wpdmpro=2014msc_thesis_michelmakepdf}, year = {2014}, date = {20140428}, school = {Technical University of Delft, the Netherlands}, abstract = {Floating wind turbines are becoming fashionable within the Renewable Energy world. In the last years MARIN has been involved in an increasing number of projects for the offshore wind industry. Model tests are often used for validating and optimizing the floater design before construction starts. A key point of model testing floating wind turbines is that wind and waves are presented simultaneously in the basin. This makes it possible to study the complex motions and interactions between the rotating turbine and the moving platform. However the experiments are done using smaller scaled models. While for the underwater loads Froude scaling laws are used successfully in the Offshore industry, the same should not be done for the aerodynamic loads. Due to the strong Reynolds scale effects, the flow regime on the blades is critical or even subcritical, and therefore laminarturbulent transition and flowseparation effects play an important role. The traditional potentialflow based tools used for design and analysis of turbines (BladeElementMomentumTheory BEMT) were not intended to work in these regimes, nor the inviscidviscous (BoundaryElementMethod BEM) tools, like XFOIL, used to obtain the turbine sections Cl/Cd/Cm input for the BEMT calculations. The complete simulation of a fullscale freefloating wind turbine under waves and winds using viscousflow (UnsteadyReynoldsAveragedNavierStokes URANS) CFD codes is still nowadays very costly, if not impossible. However these CFD theoretically more accurate methods, can be used in an efficient way for aerodynamic analysis. And they can be used rather to generate 2D input for the BEMT design tools or for the real complete analysis of the wind turbine. In the present work CFD URANS code ReFRESCO is used for both purposes, having in mind the design of the new MARIN Stock (not Floating) Wind Turbine (MSWT), based on the 5MW NREL fullscale turbine. Only openwater constant wind, fixed platform conditions are considered here. The objectives of the work presented are therefore threefold: 1) the NREL 5MW baseline turbine is calculated using ReFRESCO both in fullscale and modelscale (Froudescaling) conditions and the scaleeffects studied and quantified; 2) the MSWT designed for thrust and performancescaling is analyzed using CFD and validation against available MARIN experimental data is done; 3) in order to possibly further improve the MSWT design, the aerodynamic characteristics of its sections/foils are scrutinized by means of a full numerical study using ReFRESCO. The poor performance of the NREL 5MW turbine is due to a fully separated flow over the full range of tip speed ratios. Additionally decambering laminar separation bubbles are observed at the pressures side of the blades, further decreasing the aerodynamic performance of the turbine. Although laminar separation bubbles are not observed for the modelscale MSWT, separation does occur over the full span of the suction side of the blades. For the performancescaled MSWT, however, an attached flow region is observed at the blade tips for the higher tip speed ratios, resulting in increased CP /CT values and performance. Flow separation at fullscale conditions is present only for the heavily loaded operating conditions. These separated regions show large radial velocity components, which contradict the assumed 2D flow in BEMT models. The separated flow is also observed for the flow over the 2D airfoil sections of the MSWT. Even for small angles of attack at modelscale Reynolds numbers, separation occurs and URANS computations are necessary for larger angles of attack. For the fullscale Reynolds number regime the flow remains attached up to larger angles of attack and URANS computations are needed only for the extreme angles of attack (AoA > 14deg). The 2D flow phenomena at model and fullscale are in line with those observed for the flow over the 3D turbine. Although the MSWT has already greatly improved modelscale performance characteristics, the present research indicate that more improvements are perhaps possible. An alternative pitch angle distribution can be considered in order to reduce flow separation for even lower TSRs. Furthermore the present work showed the challenge of obtaining accurate numerical solutions for the complex unsteady flow over a wind turbine at these critical Reynolds numbers, which requires: domain studies, grid and timestep studies, good iterative convergence and an adequate turbulence model. All of these aspects were studied in this thesis.}, keywords = {BEMT, Foils, MSWT, NREL 5MW, RANS, ScaleEffects, Scaling, SpalartAllmaras, SST, Transition, Turbines, URANS, XFOIL}, pubstate = {published}, tppubtype = {mastersthesis} } Floating wind turbines are becoming fashionable within the Renewable Energy world. In the last years MARIN has been involved in an increasing number of projects for the offshore wind industry. Model tests are often used for validating and optimizing the floater design before construction starts. A key point of model testing floating wind turbines is that wind and waves are presented simultaneously in the basin. This makes it possible to study the complex motions and interactions between the rotating turbine and the moving platform. However the experiments are done using smaller scaled models. While for the underwater loads Froude scaling laws are used successfully in the Offshore industry, the same should not be done for the aerodynamic loads. Due to the strong Reynolds scale effects, the flow regime on the blades is critical or even subcritical, and therefore laminarturbulent transition and flowseparation effects play an important role. The traditional potentialflow based tools used for design and analysis of turbines (BladeElementMomentumTheory BEMT) were not intended to work in these regimes, nor the inviscidviscous (BoundaryElementMethod BEM) tools, like XFOIL, used to obtain the turbine sections Cl/Cd/Cm input for the BEMT calculations. The complete simulation of a fullscale freefloating wind turbine under waves and winds using viscousflow (UnsteadyReynoldsAveragedNavierStokes URANS) CFD codes is still nowadays very costly, if not impossible. However these CFD theoretically more accurate methods, can be used in an efficient way for aerodynamic analysis. And they can be used rather to generate 2D input for the BEMT design tools or for the real complete analysis of the wind turbine. In the present work CFD URANS code ReFRESCO is used for both purposes, having in mind the design of the new MARIN Stock (not Floating) Wind Turbine (MSWT), based on the 5MW NREL fullscale turbine. Only openwater constant wind, fixed platform conditions are considered here. The objectives of the work presented are therefore threefold: 1) the NREL 5MW baseline turbine is calculated using ReFRESCO both in fullscale and modelscale (Froudescaling) conditions and the scaleeffects studied and quantified; 2) the MSWT designed for thrust and performancescaling is analyzed using CFD and validation against available MARIN experimental data is done; 3) in order to possibly further improve the MSWT design, the aerodynamic characteristics of its sections/foils are scrutinized by means of a full numerical study using ReFRESCO. The poor performance of the NREL 5MW turbine is due to a fully separated flow over the full range of tip speed ratios. Additionally decambering laminar separation bubbles are observed at the pressures side of the blades, further decreasing the aerodynamic performance of the turbine. Although laminar separation bubbles are not observed for the modelscale MSWT, separation does occur over the full span of the suction side of the blades. For the performancescaled MSWT, however, an attached flow region is observed at the blade tips for the higher tip speed ratios, resulting in increased CP /CT values and performance. Flow separation at fullscale conditions is present only for the heavily loaded operating conditions. These separated regions show large radial velocity components, which contradict the assumed 2D flow in BEMT models. The separated flow is also observed for the flow over the 2D airfoil sections of the MSWT. Even for small angles of attack at modelscale Reynolds numbers, separation occurs and URANS computations are necessary for larger angles of attack. For the fullscale Reynolds number regime the flow remains attached up to larger angles of attack and URANS computations are needed only for the extreme angles of attack (AoA > 14deg). The 2D flow phenomena at model and fullscale are in line with those observed for the flow over the 3D turbine. Although the MSWT has already greatly improved modelscale performance characteristics, the present research indicate that more improvements are perhaps possible. An alternative pitch angle distribution can be considered in order to reduce flow separation for even lower TSRs. Furthermore the present work showed the challenge of obtaining accurate numerical solutions for the complex unsteady flow over a wind turbine at these critical Reynolds numbers, which requires: domain studies, grid and timestep studies, good iterative convergence and an adequate turbulence model. All of these aspects were studied in this thesis.  
2013 

Kuin, Roderick HYDRODYNAMIC IMPROVEMENTS OF A GENERIC SUBMARINE USING VISCOUS FLOW CALCULATIONS Masters Thesis University of Twente, Enschede, the Netherlands, 2013. Abstract  Links  BibTeX  Tags: Design, Drift, Manoeuvring, RANS, Rotation, SST, Submarines, Validation, Verification @mastersthesis{2013Msc_Thesis_RoderickKuin, title = {HYDRODYNAMIC IMPROVEMENTS OF A GENERIC SUBMARINE USING VISCOUS FLOW CALCULATIONS}, author = {Roderick Kuin}, url = {http://www.refresco.org/?wpdmpro=2013msc_thesis_roderickkuinpdf}, year = {2013}, date = {20131231}, school = {University of Twente, Enschede, the Netherlands}, abstract = {Generally, underwater vehicles such as gliders or submarines have a hull or fuselage shape with low drag properties. However, additional appendages are generally required for control or storage of equipment. These appendages induce additional resistance and may be detrimental to the quality of the inflow to the aft control surfaces or propeller. This, in turn, can lead to loss of propulsion performance or increase of vibrations and radiated noise. The underlying hydrodynamic mechanism is the penetration by the appendage of the boundary layer developing on the hull, which causes the formation of a socalled horseshoe vortex in a region of separated flow near the stagnation area on the appendage. Computational Fluid Dynamics (CFD) has matured to a state that it can be applied successfully to investigate and optimise the flow around ships and offshore structures. In this research, CFD is used to study the flow around a typical wingbody junction in order to obtain insight in how to suppress the horseshoe vortex that is wrapped around the appendage. A generic submarine hull shape has been selected and the impact of a range of modifications of the sail (sometimes called fin or fairwater) on the resistance, propulsion, manoeuvring and wake field have been investigated. To quantify the nonuniformity of the wake field, a socalled Wake Object Function (WOF) is used. The WOF is defined such that decreasing its value reduces the chance of (erosive) cavitation and radiated noise. This research presents the results of the CFD computations for a number of sail variants and discusses the changes in the flow in detail. Design guidelines regarding the most promising modifications have been developed. It is shown that a thicker and a tapered sail have a significant negative influence on the main hydrodynamic characteristics, however, quite significant improvements of the resistance as well as the wake quality can be obtained by properly designing the junction between the sail and the hull.}, keywords = {Design, Drift, Manoeuvring, RANS, Rotation, SST, Submarines, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } Generally, underwater vehicles such as gliders or submarines have a hull or fuselage shape with low drag properties. However, additional appendages are generally required for control or storage of equipment. These appendages induce additional resistance and may be detrimental to the quality of the inflow to the aft control surfaces or propeller. This, in turn, can lead to loss of propulsion performance or increase of vibrations and radiated noise. The underlying hydrodynamic mechanism is the penetration by the appendage of the boundary layer developing on the hull, which causes the formation of a socalled horseshoe vortex in a region of separated flow near the stagnation area on the appendage. Computational Fluid Dynamics (CFD) has matured to a state that it can be applied successfully to investigate and optimise the flow around ships and offshore structures. In this research, CFD is used to study the flow around a typical wingbody junction in order to obtain insight in how to suppress the horseshoe vortex that is wrapped around the appendage. A generic submarine hull shape has been selected and the impact of a range of modifications of the sail (sometimes called fin or fairwater) on the resistance, propulsion, manoeuvring and wake field have been investigated. To quantify the nonuniformity of the wake field, a socalled Wake Object Function (WOF) is used. The WOF is defined such that decreasing its value reduces the chance of (erosive) cavitation and radiated noise. This research presents the results of the CFD computations for a number of sail variants and discusses the changes in the flow in detail. Design guidelines regarding the most promising modifications have been developed. It is shown that a thicker and a tapered sail have a significant negative influence on the main hydrodynamic characteristics, however, quite significant improvements of the resistance as well as the wake quality can be obtained by properly designing the junction between the sail and the hull.  
Willemsen, Christiaan Improving Potential Flow Predictions for Ducted Propellers Masters Thesis University of Twente, Enschede, the Netherlands, 2013. Abstract  Links  BibTeX  Tags: Ducts, Propeller, RANS, RANSBEM Coupling, SST, Validation, Verification @mastersthesis{2013Msc_Thesis_ChrisWillemsen, title = {Improving Potential Flow Predictions for Ducted Propellers}, author = {Christiaan Willemsen}, url = {http://www.refresco.org/?wpdmpro=2013msc_thesis_chriswillemsenpdf}, year = {2013}, date = {20131213}, school = {University of Twente, Enschede, the Netherlands}, abstract = {The advantages of propulsion systems using a thrust generating duct around a propeller are well known in naval architecture. A ducted propeller is often employed to increase the efficiency and thrust of a highly loaded propeller. The flow accelerating duct can contribute to 50 % of the propulsor total thrust at zero ship speed. There are not many fast and accurate hydrodynamic prediction methods for the design phase of ducted propellers. Model tests are expensive, while computations based on the Reynoldsaveraged NavierStokes (RANS) equations require long CPU times. Therefore these approaches are not yet routinely used in the design process of propulsors. Currently the design process is mostly based on the use of potential flow methods, like the MARIN Boundary Element Method (BEM) PROCAL. This method is efficient and is able to deliver accurate predictions of the forces acting on open propellers, but it is less accurate when viscous flow effects become important such as is the case for ducted propellers. The goal of the present research is to investigate the flow around a ducted propeller using the MARIN inhousedeveloped RANS method ReFRESCO, with particular emphasis on the in influence of the viscous flow effects such as boundary layers, tip vortices and flow separation on the outer surface of the duct. The results obtained with RANS are used to improve the prediction of PROCAL. Finally a coupling between PROCAL and ReFRESCO is accomplished to include the viscous flow effects in an efficient way. The viscous flow over the duct is annalyzed using ReFRESCO, in which the propeller action is represented by body forces added to the righthandside of the momentum equations. These body forces are obtained from a PROCAL computation for the ducted propeller in which the propeller is represented as a discrete number of blades. Results of this approach show a good agreement between experiments and the numerical simulations: the forces differ less than 1 % around the design point of the ducted propeller.}, keywords = {Ducts, Propeller, RANS, RANSBEM Coupling, SST, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } The advantages of propulsion systems using a thrust generating duct around a propeller are well known in naval architecture. A ducted propeller is often employed to increase the efficiency and thrust of a highly loaded propeller. The flow accelerating duct can contribute to 50 % of the propulsor total thrust at zero ship speed. There are not many fast and accurate hydrodynamic prediction methods for the design phase of ducted propellers. Model tests are expensive, while computations based on the Reynoldsaveraged NavierStokes (RANS) equations require long CPU times. Therefore these approaches are not yet routinely used in the design process of propulsors. Currently the design process is mostly based on the use of potential flow methods, like the MARIN Boundary Element Method (BEM) PROCAL. This method is efficient and is able to deliver accurate predictions of the forces acting on open propellers, but it is less accurate when viscous flow effects become important such as is the case for ducted propellers. The goal of the present research is to investigate the flow around a ducted propeller using the MARIN inhousedeveloped RANS method ReFRESCO, with particular emphasis on the in influence of the viscous flow effects such as boundary layers, tip vortices and flow separation on the outer surface of the duct. The results obtained with RANS are used to improve the prediction of PROCAL. Finally a coupling between PROCAL and ReFRESCO is accomplished to include the viscous flow effects in an efficient way. The viscous flow over the duct is annalyzed using ReFRESCO, in which the propeller action is represented by body forces added to the righthandside of the momentum equations. These body forces are obtained from a PROCAL computation for the ducted propeller in which the propeller is represented as a discrete number of blades. Results of this approach show a good agreement between experiments and the numerical simulations: the forces differ less than 1 % around the design point of the ducted propeller.  
2012 

Kamphuis, Nico ANALYSIS OF REFRESCO COMPUTATIONS APPLIED TO A TIP VORTEX OF A WING Masters Thesis University of Twente, Enschede, the Netherlands, 2012. Links  BibTeX  Tags: Chow Wing, EARSM, RANS, SpalartAllmaras, SST, Tipvortex, Turbulence Models @mastersthesis{2011Stage_NicoKamphuis, title = {ANALYSIS OF REFRESCO COMPUTATIONS APPLIED TO A TIP VORTEX OF A WING}, author = {Nico Kamphuis}, url = {http://www.refresco.org/?wpdmpro=2011stage_nicokamphuispdf}, year = {2012}, date = {20121201}, school = {University of Twente, Enschede, the Netherlands}, keywords = {Chow Wing, EARSM, RANS, SpalartAllmaras, SST, Tipvortex, Turbulence Models}, pubstate = {published}, tppubtype = {mastersthesis} }  
Pereira, Filipe Verication of ReFRESCO with the Method of Manufactured Solutions Masters Thesis IST, Lisbon, Portugal, 2012. Abstract  Links  BibTeX  Tags: Code Verification, Convection schemes, Excentricity, MMS, Nonorthogonality, RANS, SST, Validation, Verification @mastersthesis{2012Msc_Thesis_FilipePereira, title = {Verication of ReFRESCO with the Method of Manufactured Solutions}, author = {Filipe Pereira}, url = {http://www.refresco.org/?wpdmpro=2012msc_thesis_filipepereirapdf}, year = {2012}, date = {20121001}, school = {IST, Lisbon, Portugal}, abstract = {The purpose of this Thesis was to Verify the RANS solver ReFRESCO. This analysis was executed over three distinct parts of the code: convection schemes, nonorthogonality and excentricity correctors. Moreover, it was performed the implementation and evaluation of the numerical properties of two nonorthogonality and three excentricity new correction methods. In order to execute the Verification of ReFRESCO, grid refinement studies were performed to check if the numerical error tend to zero with the correct order of grid convergence (theoretical order). The calculation of the numerical error required the use of the Method of Manufactured Solutions to create exact solutions of the RANS equations. Thus, three manufactured solutions were used, each one resembling a different flow. The main conclusions of the present Thesis were: the convection schemes are correctly coded; the Hybrid scheme order of grid convergence did not vary linearly with the blending factor and it tended to a step function with the increase of the ow complexity; the tests performed over the nonorthogonality correctors showed that these methods maintained the secondorder of the code while discarding the correctors originated a constant numerical error in the solution; in grids where the excentricity factor was independent from grid renement, compared to the noncorrected case (constant numerical error), the excentricity correctors (correctly implemented) decreased significantly the magnitude of the numerical error. However, these correctors only guaranteed rstorder accuracy.}, keywords = {Code Verification, Convection schemes, Excentricity, MMS, Nonorthogonality, RANS, SST, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } The purpose of this Thesis was to Verify the RANS solver ReFRESCO. This analysis was executed over three distinct parts of the code: convection schemes, nonorthogonality and excentricity correctors. Moreover, it was performed the implementation and evaluation of the numerical properties of two nonorthogonality and three excentricity new correction methods. In order to execute the Verification of ReFRESCO, grid refinement studies were performed to check if the numerical error tend to zero with the correct order of grid convergence (theoretical order). The calculation of the numerical error required the use of the Method of Manufactured Solutions to create exact solutions of the RANS equations. Thus, three manufactured solutions were used, each one resembling a different flow. The main conclusions of the present Thesis were: the convection schemes are correctly coded; the Hybrid scheme order of grid convergence did not vary linearly with the blending factor and it tended to a step function with the increase of the ow complexity; the tests performed over the nonorthogonality correctors showed that these methods maintained the secondorder of the code while discarding the correctors originated a constant numerical error in the solution; in grids where the excentricity factor was independent from grid renement, compared to the noncorrected case (constant numerical error), the excentricity correctors (correctly implemented) decreased significantly the magnitude of the numerical error. However, these correctors only guaranteed rstorder accuracy.  
2011 

Otto, William NUMERICAL SIMULATIONS OF FLOW OVER AN AXIAL MARINE CURRENT TURBINE Masters Thesis Technical University of Delft, the Netherlands, 2011. Abstract  Links  BibTeX  Tags: Current Turbines, RANS, SST, Turbines, URANS, Validation, Verification @mastersthesis{2011Msc_Thesis_WilliamOtto, title = {NUMERICAL SIMULATIONS OF FLOW OVER AN AXIAL MARINE CURRENT TURBINE}, author = {William Otto}, url = {http://www.refresco.org/?wpdmpro=2011msc_thesis_williamottopdf}, year = {2011}, date = {20111011}, school = {Technical University of Delft, the Netherlands}, abstract = {The main objective of this Msc. thesis is to obtain and analyze numerical simulations of singlephase flow over an axial marine current turbine. A wide range of operating conditions is simulated. Great attention is paid to verification, validation and uncertainty analysis. As benchmark, a reference turbine with experimental data is used which is found in literature (A.S. Bahaj and W.M.J. Batten, 2005 [17]). The simulations were performed at model scale and scale effects were studied by using the same geometry at full scale Reynolds numbers. This thesis is limited to single phase flows, what means that cavitation and free surface effects are deliberately excluded. Only a uniform inflow is modeled and interaction between the turbine and other objects as walls, floors, mounting rigs or other turbines are not taken into account (’open water condition’). Because these aspects can play a significant roll in practical applications, the numerical method is chosen such that they can be implemented in future work, once verified and validated simulations of noninteracting, singlephase flow have been obtained. Because its ability to include the aforementioned effects, as well its the ability to study scale effects, the MARIN inhouse RANS solver ReFRESCO is used for the simulations. A geometrical description of the reference turbine was received from the original authors. This geometry is modified in order to obtain feasible calculations. First, the trailing edge had to be thickened in order to avoid troubles in the grid generation. Second, a new connection has been constructed between the blades and the hub. The original connection causes an unsteady wake which elongates the calculation time to weeks. With a new constructed blade to hub connection, the flow is less complex, reducing the calculation time to a couple of days per condition. The modeling error caused by the thickened trailing edge is studied by using two dimensional RANS calculations over a radial section of the turbine (r=R = 0:7). It is estimated that the sectional lift is reduced by 3.78% due to the thickened trailing edge. Also an increase in drag is obtained, which is estimated as 6.35%. The turbine power and axial loading is corrected for this effect. The modified blade to hub connection is taken into account as an additional uncertainty in the solutions. A verification and validation procedure is performed to estimate the numerical and modeling uncertainties. The largest component of the numerical uncertainty is the discretization error. This error is hard to quantify due to: 1) the unstructured grid approach what makes it hard to produce a series of geometrical similar grids, 2) the small refinement range limited by the available memory resources. Therefore, a conservative estimation is made by using a safety factor. The numerical uncertainty is estimated as U = 3:6% for the power coefficient CP and U = 4:8% for the axial loading coefficient CT . A cylindrical computational domain is used to represent the open water condition. Initially, the domain size was 8 turbine diameter wide in radial direction. Later it proved that this domain was too small to fully represent an undisturbed flow without (numerical) blockage effects. By systematically increasing the domain size, it is estimated that the modeling error caused by the too small domain is Udomain = 0:5% for CP and Udomain = 2:6% for CT . The calculation results at model scale (Re = 1:4 105) show a very good similarity with the experimental results for the power production as well as the axial loading. Due to the scatter in the experiments, it is not possible to follow an official validation procedure. The flow analysis at model scale shows a large area of laminar flow separation at the suction side of the blades. It can be said that the blades are in stall for a large part. The turbulence intensity shows the boundary layer at the blade is in the transitional region. Roughly half of the chord length has a laminar boundary layer, the second half is turbulent. The stall can be caused by the laminar boundary layer, what makes it a scale effect. The flow analysis at full scale Reynolds numbers Re = 5 106 does not show the large separation areas. A fully turbulent boundary layer is obtained and the flow stays to a great extend attached to the blade. As a consequence, the obtained axial loading and power coefficient is more than 10% higher than at model scale. This is a significant scale effect where designers of marine current turbines should be aware of.}, keywords = {Current Turbines, RANS, SST, Turbines, URANS, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } The main objective of this Msc. thesis is to obtain and analyze numerical simulations of singlephase flow over an axial marine current turbine. A wide range of operating conditions is simulated. Great attention is paid to verification, validation and uncertainty analysis. As benchmark, a reference turbine with experimental data is used which is found in literature (A.S. Bahaj and W.M.J. Batten, 2005 [17]). The simulations were performed at model scale and scale effects were studied by using the same geometry at full scale Reynolds numbers. This thesis is limited to single phase flows, what means that cavitation and free surface effects are deliberately excluded. Only a uniform inflow is modeled and interaction between the turbine and other objects as walls, floors, mounting rigs or other turbines are not taken into account (’open water condition’). Because these aspects can play a significant roll in practical applications, the numerical method is chosen such that they can be implemented in future work, once verified and validated simulations of noninteracting, singlephase flow have been obtained. Because its ability to include the aforementioned effects, as well its the ability to study scale effects, the MARIN inhouse RANS solver ReFRESCO is used for the simulations. A geometrical description of the reference turbine was received from the original authors. This geometry is modified in order to obtain feasible calculations. First, the trailing edge had to be thickened in order to avoid troubles in the grid generation. Second, a new connection has been constructed between the blades and the hub. The original connection causes an unsteady wake which elongates the calculation time to weeks. With a new constructed blade to hub connection, the flow is less complex, reducing the calculation time to a couple of days per condition. The modeling error caused by the thickened trailing edge is studied by using two dimensional RANS calculations over a radial section of the turbine (r=R = 0:7). It is estimated that the sectional lift is reduced by 3.78% due to the thickened trailing edge. Also an increase in drag is obtained, which is estimated as 6.35%. The turbine power and axial loading is corrected for this effect. The modified blade to hub connection is taken into account as an additional uncertainty in the solutions. A verification and validation procedure is performed to estimate the numerical and modeling uncertainties. The largest component of the numerical uncertainty is the discretization error. This error is hard to quantify due to: 1) the unstructured grid approach what makes it hard to produce a series of geometrical similar grids, 2) the small refinement range limited by the available memory resources. Therefore, a conservative estimation is made by using a safety factor. The numerical uncertainty is estimated as U = 3:6% for the power coefficient CP and U = 4:8% for the axial loading coefficient CT . A cylindrical computational domain is used to represent the open water condition. Initially, the domain size was 8 turbine diameter wide in radial direction. Later it proved that this domain was too small to fully represent an undisturbed flow without (numerical) blockage effects. By systematically increasing the domain size, it is estimated that the modeling error caused by the too small domain is Udomain = 0:5% for CP and Udomain = 2:6% for CT . The calculation results at model scale (Re = 1:4 105) show a very good similarity with the experimental results for the power production as well as the axial loading. Due to the scatter in the experiments, it is not possible to follow an official validation procedure. The flow analysis at model scale shows a large area of laminar flow separation at the suction side of the blades. It can be said that the blades are in stall for a large part. The turbulence intensity shows the boundary layer at the blade is in the transitional region. Roughly half of the chord length has a laminar boundary layer, the second half is turbulent. The stall can be caused by the laminar boundary layer, what makes it a scale effect. The flow analysis at full scale Reynolds numbers Re = 5 106 does not show the large separation areas. A fully turbulent boundary layer is obtained and the flow stays to a great extend attached to the blade. As a consequence, the obtained axial loading and power coefficient is more than 10% higher than at model scale. This is a significant scale effect where designers of marine current turbines should be aware of.  
Toxopeus, Serge Practical application of viscousflow calculations for the simulation of manoeuvring ships PhD Thesis Technical University of Delft, the Netherlands, 2011, ISBN: ISBN 9789075757057. Abstract  Links  BibTeX  Tags: DARPA Suboff, Drag, DTMB 5415M, Esso Osaka, HTC, KCS, KVLCC2, Lift, Manoeuvring, RANS, Rotation, Series60, Ships, SST, Submarines, Validation, Verification, Yaw @phdthesis{2011PhDToxopeus, title = {Practical application of viscousflow calculations for the simulation of manoeuvring ships}, author = {Serge Toxopeus}, url = {http://www.refresco.org/?wpdmpro=2011phdtoxopeuspdf}, isbn = {ISBN 9789075757057}, year = {2011}, date = {20110509}, school = {Technical University of Delft, the Netherlands}, abstract = {The present work was initiated in order to improve traditional manoeuvring simulations based on empirical equations to model the forces and moments on the ship. With the evolution of the capability of viscousflow solvers to predict forces and moments on ships, it was decided to develop a practical method to simulate the manoeuvrability of ships in which viscousflow solvers are utilised and to investigate whether this improves the accuracy of manoeuvring predictions. To achieve this goal, the virtual captive test approach is adopted, because of the efficient use of computational resources compared to other methods. This procedure mimics the approach for manoeuvring simulations in which experimental PMM is used to obtain the forces and moments on the ship. This study extends the work of other researchers by providing extensive verification and validation of the predicted forces and moments on the hull and a detailed study of the sensitivity of the manoeuvring characteristics of the ship to changes in the hydrodynamic coefficients in the simulation model. Changes in the flow solvers were required to be able to calculate the flow around ships in rotational motion. These changes are discussed as well as the acceleration techniques that were developed to reduce the effort spent on grid generation and during the computations. In this thesis, it is demonstrated that good predictions of the loads on the hull in manoeuvring motion can be obtained for a wide range of ship types. The trends in the forces and moments as a function of the drift angle or yaw rate are simulated well. The verification studies provide useful insight into the in influence of grid density on the predicted forces and moments. In several cases, validation of the calculations failed, indicating modelling errors in the numerical results. In these cases, it was generally seen that the magnitude of the transverse force was underpredicted, while the magnitude of the yaw moment was overpredicted. For manoeuvring studies in the early design, the comparison errors are within acceptable levels. However, improvements remain desired and may be obtained using finer grids, larger domain sizes, different grid topologies with refinement in the wake of the ship, other turbulence models or incorporating free surface deformation. The manoeuvring prediction program SurSim has been used to simulate the manoeuvrability of the HTC. A procedure is proposed to derive the hydrodynamic coefficients required to model the forces and moments on the bare hull. This procedure is chosen to enable accurate modelling of the linearised behaviour for coursekeeping as well as realistic modelling of the harbour manoeuvring characteristics, and to enable the modelling of nonlinear manoeuvres accurately. To generate validation data for the manoeuvring predictions presented in this thesis, free sailing manoeuvring tests for the HTC were performed. This test campaign resulted in a very valuable data set which can be used for public validation studies. Besides obtaining general characteristics of the manoeuvrability of a singlescrew container ship, unique information has been obtained on the drift angles and rates of turn combined with propeller and rudder forces. Furthermore, repeat tests have been conducted for selected manoeuvres. Based on these tests, the uncertainty in the characteristic manoeuvring properties has been estimated. By using hydrodynamic manoeuvring coefficients derived from the CFD calculations, it has been shown that it is possible to improve the prediction of ship manoeuvres compared to predictions using coefficients based on empirical equations. A considerable improvement in the turning circle predictions was obtained. The prediction of the yaw checking and course keeping and initial turning abilities based on zigzag simulations improved as well, but further improvements are required for more reliable assessment of the manoeuvring performance. The sensitivity of the manoeuvring predictions to changes in the hydrodynamic coefficients was studied. It was found that for accurate predictions of the manoeuvrability using coefficients derived from CFD calculations, accurate predictions of especially the yawing moment must be made.}, keywords = {DARPA Suboff, Drag, DTMB 5415M, Esso Osaka, HTC, KCS, KVLCC2, Lift, Manoeuvring, RANS, Rotation, Series60, Ships, SST, Submarines, Validation, Verification, Yaw}, pubstate = {published}, tppubtype = {phdthesis} } The present work was initiated in order to improve traditional manoeuvring simulations based on empirical equations to model the forces and moments on the ship. With the evolution of the capability of viscousflow solvers to predict forces and moments on ships, it was decided to develop a practical method to simulate the manoeuvrability of ships in which viscousflow solvers are utilised and to investigate whether this improves the accuracy of manoeuvring predictions. To achieve this goal, the virtual captive test approach is adopted, because of the efficient use of computational resources compared to other methods. This procedure mimics the approach for manoeuvring simulations in which experimental PMM is used to obtain the forces and moments on the ship. This study extends the work of other researchers by providing extensive verification and validation of the predicted forces and moments on the hull and a detailed study of the sensitivity of the manoeuvring characteristics of the ship to changes in the hydrodynamic coefficients in the simulation model. Changes in the flow solvers were required to be able to calculate the flow around ships in rotational motion. These changes are discussed as well as the acceleration techniques that were developed to reduce the effort spent on grid generation and during the computations. In this thesis, it is demonstrated that good predictions of the loads on the hull in manoeuvring motion can be obtained for a wide range of ship types. The trends in the forces and moments as a function of the drift angle or yaw rate are simulated well. The verification studies provide useful insight into the in influence of grid density on the predicted forces and moments. In several cases, validation of the calculations failed, indicating modelling errors in the numerical results. In these cases, it was generally seen that the magnitude of the transverse force was underpredicted, while the magnitude of the yaw moment was overpredicted. For manoeuvring studies in the early design, the comparison errors are within acceptable levels. However, improvements remain desired and may be obtained using finer grids, larger domain sizes, different grid topologies with refinement in the wake of the ship, other turbulence models or incorporating free surface deformation. The manoeuvring prediction program SurSim has been used to simulate the manoeuvrability of the HTC. A procedure is proposed to derive the hydrodynamic coefficients required to model the forces and moments on the bare hull. This procedure is chosen to enable accurate modelling of the linearised behaviour for coursekeeping as well as realistic modelling of the harbour manoeuvring characteristics, and to enable the modelling of nonlinear manoeuvres accurately. To generate validation data for the manoeuvring predictions presented in this thesis, free sailing manoeuvring tests for the HTC were performed. This test campaign resulted in a very valuable data set which can be used for public validation studies. Besides obtaining general characteristics of the manoeuvrability of a singlescrew container ship, unique information has been obtained on the drift angles and rates of turn combined with propeller and rudder forces. Furthermore, repeat tests have been conducted for selected manoeuvres. Based on these tests, the uncertainty in the characteristic manoeuvring properties has been estimated. By using hydrodynamic manoeuvring coefficients derived from the CFD calculations, it has been shown that it is possible to improve the prediction of ship manoeuvres compared to predictions using coefficients based on empirical equations. A considerable improvement in the turning circle predictions was obtained. The prediction of the yaw checking and course keeping and initial turning abilities based on zigzag simulations improved as well, but further improvements are required for more reliable assessment of the manoeuvring performance. The sensitivity of the manoeuvring predictions to changes in the hydrodynamic coefficients was studied. It was found that for accurate predictions of the manoeuvrability using coefficients derived from CFD calculations, accurate predictions of especially the yawing moment must be made.  
Peyro, Guillaume Analysis of Flows on Stabilizer Fins using ReFRESCO: 2D,3D, Static and Dynamic Eects Masters Thesis ENSTA, Brest, Bretagne, France, 2011. Abstract  Links  BibTeX  Tags: Imposed Motion, NACA 0015, RANS, SST, Stabilizer fins, URANS, Validation, Verification @mastersthesis{2011Msc_Thesis_GuilaumePeyro, title = {Analysis of Flows on Stabilizer Fins using ReFRESCO: 2D,3D, Static and Dynamic Eects}, author = {Guillaume Peyro}, url = { http://www.refresco.org/?wpdmpro=2011msc_thesis_guilaumepeyropdf}, year = {2011}, date = {20110301}, school = {ENSTA, Brest, Bretagne, France}, abstract = {Nowadays, Computational Fluid Dynamic (CFD) is becoming more and more important in the maritime field to investigate complex hydrodynamic phenomena, especially when combined with model tests. Conscious about this, the Maritime Research Institute of Netherlands (MARIN) developed its own CFD code called ReFRESCO. The aim of this study is to use CFD to investigate the flow on a NACA 0015 hydrofoil which represents a stabilizer n, and to compare CFD results with model tests of this n. For this study, first, 2D static computations are done with ReFRESCO which allows us to know the effects of numerical parameters (boundary conditions, domain dimensions, domain shape...) on the results. Using these results, similar 3D computations are performed and compared to the results of the model tests. Finally, preliminary dynamic computations are done in order to test some tools simulating the oscillation of the foil.}, keywords = {Imposed Motion, NACA 0015, RANS, SST, Stabilizer fins, URANS, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } Nowadays, Computational Fluid Dynamic (CFD) is becoming more and more important in the maritime field to investigate complex hydrodynamic phenomena, especially when combined with model tests. Conscious about this, the Maritime Research Institute of Netherlands (MARIN) developed its own CFD code called ReFRESCO. The aim of this study is to use CFD to investigate the flow on a NACA 0015 hydrofoil which represents a stabilizer n, and to compare CFD results with model tests of this n. For this study, first, 2D static computations are done with ReFRESCO which allows us to know the effects of numerical parameters (boundary conditions, domain dimensions, domain shape...) on the results. Using these results, similar 3D computations are performed and compared to the results of the model tests. Finally, preliminary dynamic computations are done in order to test some tools simulating the oscillation of the foil.  
2009 

Delvoye, Simon Simulation and analysis of the flow around an underwater exhaust with FreSCo Masters Thesis ISITV, Toulon, France, 2009. Links  BibTeX  Tags: Design, Multiphase, RANS, Scoops, SpalartAllmaras, SST, URANS, Verification @mastersthesis{2009Msc_Thesis_SimonDelvoye, title = {Simulation and analysis of the flow around an underwater exhaust with FreSCo}, author = {Delvoye, Simon}, url = {http://www.refresco.org/?wpdmpro=2009msc_thesis_simondelvoyepdf }, year = {2009}, date = {20090807}, school = {ISITV, Toulon, France}, keywords = {Design, Multiphase, RANS, Scoops, SpalartAllmaras, SST, URANS, Verification}, pubstate = {published}, tppubtype = {mastersthesis} }  
2008 

Rijpkema, Douwe Numerical Simulation of SinglePhase and MultiPhase Flow over a NACA 0015 Hydrofoil Masters Thesis Technical University of Delft, the Netherlands, 2008. Abstract  Links  BibTeX  Tags: Cavitation, Drag, Lift, NACA 0015, RANS, URANS, Validation, Verification @mastersthesis{2008Msc_Thesis_DouweRijpkema, title = {Numerical Simulation of SinglePhase and MultiPhase Flow over a NACA 0015 Hydrofoil}, author = {Douwe Rijpkema}, url = {http://www.refresco.org/?wpdmpro=2008msc_thesis_douwerijpkemapdf}, year = {2008}, date = {20081107}, school = {Technical University of Delft, the Netherlands}, abstract = {In the design of marine propellers, cavitation  the phenomenon of vapour formation due to a pressure reduction at constant temperature  is associated with negative effects on the performance and lifespan of the propeller. Additionally cavitation can be a source of inboard and underwater noise. Therefore insight in the occurence of cavitation and the development of the cavity on a propeller is essential. The numerical simulation of this phenomenon with computational fluid dynamics (CFD) tools may play an important role in this analysis. In this study a numerical simulation of both wetted and cavitating flow over a NACA 0015 hydrofoil is performed. The foil is placed at an angle of attack of 6 degrees for a Reynolds number of 1.5E6. The CFD code FreSCo is used for the numerical simulations. FreSCo is an unsteady RANS solver actively being developed by a cooperation of Maritime Research Institute Netherlands (MARIN), Hamburgische SchiffbauVersuchsanstalt (HSVA) and Technische Universitaet HamburgHarburg (TUHH). In the computations, a MenterSST turbulence model is applied and a volume of fluid approach is used for the modelling of multiple phases. The influence of various numerical parameters on the hydrodynamic forces and cavitation behaviour is investigated. For wetted flow a variation in grid topology showed a significant influence on the lift. Computations with the Otype grid resulted in an increase in lift in comparison to the Ctype grid results. The Otype grid showed a better agreement with the pressure distributions obtained by other numerical methods. A refinement of the grid produced less variation in results for the QUICK scheme compared to the blending scheme for the convective flux term in the momentum equations. Therefore an Otype grid combined with a QUICK convection scheme is preferred for this type of flow. The comparison of the pressure distributions between FreSCo and two different boundary element methods showed a good agreement in results. In the case of cavitating flow, the cavitation model accounts for the creation and destruction of the vapour in the liquid. It gives an expression for the source term in the transport equation of the vapour volume fraction. A comparison was made between the Sauer, Zwart and Kunz cavitation models. The different formulations for the source term of the three models, led to large deviations in results and affected the numerical stability. It was observed that for the same cavitation model a reduction in tuning coefficient resulted in a smaller cavity and more numerically stable behaviour. For the high cavitation numbers ( = 1.75, = 1.5 and = 1.25) a steady attached cavity was found on the foil for all cavitation models. The different formulation of the various cavitation models resulted in a difference in hydrodynamic forces and cavity characteristics, becoming more pronounced with decreasing cavitation number. Due to the large source terms of the cavitation models, large pressure oscillations and consequently large lift and drag peaks were observed at low cavitation numbers ( = 1.0). A reduction of the timestep or of the condensation tuning coefficient resulted in a more stable computation. For = 1.0 shedding of the cavity was observed in the initial phase of the computation, eventually leading to an attached cavity on the foil that periodically varied in size}, keywords = {Cavitation, Drag, Lift, NACA 0015, RANS, URANS, Validation, Verification}, pubstate = {published}, tppubtype = {mastersthesis} } In the design of marine propellers, cavitation  the phenomenon of vapour formation due to a pressure reduction at constant temperature  is associated with negative effects on the performance and lifespan of the propeller. Additionally cavitation can be a source of inboard and underwater noise. Therefore insight in the occurence of cavitation and the development of the cavity on a propeller is essential. The numerical simulation of this phenomenon with computational fluid dynamics (CFD) tools may play an important role in this analysis. In this study a numerical simulation of both wetted and cavitating flow over a NACA 0015 hydrofoil is performed. The foil is placed at an angle of attack of 6 degrees for a Reynolds number of 1.5E6. The CFD code FreSCo is used for the numerical simulations. FreSCo is an unsteady RANS solver actively being developed by a cooperation of Maritime Research Institute Netherlands (MARIN), Hamburgische SchiffbauVersuchsanstalt (HSVA) and Technische Universitaet HamburgHarburg (TUHH). In the computations, a MenterSST turbulence model is applied and a volume of fluid approach is used for the modelling of multiple phases. The influence of various numerical parameters on the hydrodynamic forces and cavitation behaviour is investigated. For wetted flow a variation in grid topology showed a significant influence on the lift. Computations with the Otype grid resulted in an increase in lift in comparison to the Ctype grid results. The Otype grid showed a better agreement with the pressure distributions obtained by other numerical methods. A refinement of the grid produced less variation in results for the QUICK scheme compared to the blending scheme for the convective flux term in the momentum equations. Therefore an Otype grid combined with a QUICK convection scheme is preferred for this type of flow. The comparison of the pressure distributions between FreSCo and two different boundary element methods showed a good agreement in results. In the case of cavitating flow, the cavitation model accounts for the creation and destruction of the vapour in the liquid. It gives an expression for the source term in the transport equation of the vapour volume fraction. A comparison was made between the Sauer, Zwart and Kunz cavitation models. The different formulations for the source term of the three models, led to large deviations in results and affected the numerical stability. It was observed that for the same cavitation model a reduction in tuning coefficient resulted in a smaller cavity and more numerically stable behaviour. For the high cavitation numbers ( = 1.75, = 1.5 and = 1.25) a steady attached cavity was found on the foil for all cavitation models. The different formulation of the various cavitation models resulted in a difference in hydrodynamic forces and cavity characteristics, becoming more pronounced with decreasing cavitation number. Due to the large source terms of the cavitation models, large pressure oscillations and consequently large lift and drag peaks were observed at low cavitation numbers ( = 1.0). A reduction of the timestep or of the condensation tuning coefficient resulted in a more stable computation. For = 1.0 shedding of the cavity was observed in the initial phase of the computation, eventually leading to an attached cavity on the foil that periodically varied in size 
ReFRESCO related MSc and Phd ThesesDennis van Espelo20190619T14:01:23+01:00