Abstract
While historically sailmaking and saildesign were considered as arts, in the 20th century, mainly from the 1980s onwards, engineering sciences have started to play an important role. Two fields are of particular interest: structural and fluid mechanics. Initially, the sails were tested in the wind tunnel, aggregate flow forces measured and the interaction of flow and structural behaviour implicitly captured by visual observation. No quantitative structural assessment was available in these experiments. With the advent of affordable powerful personal computers, programs were developed to compute the flow around sails and the structural reaction to the resulting forces. These programs were based on significantly simplified assumptions about the fluid mechanics  potential flow  as well as the complete neglect of any unsteady behaviour of flow or coupled result. These simplifications limit the applicability of these programs to upwind sails, essentially this airfoils working at small angles of attack. As downwind sails do not comply with these limitations they are still tested in the wind tunnel with the associated scale effects and limited outcome of quantitative results.
Within this thesis a method is being developed to capture the interaction between the complex viscous flow around downwind sails and compute the structural answer to the resulting forces. First a structural model suitable for downwind sails is developed. This is coupled to a commercial solver for simulations of viscous flow. The individual parts (structural and flow simulation as well as coupling) and the entire method are verified and validated. Finally an application example is given. First, the structural model and coupling to the flow solver are developed. The particular challenge regarding the structural model is the requirement to compute the complex behaviour of downwind sails. By design these sails have negligible bending stiffness with the material being stiff in tension but without any meaningful compressive stiffness. To this end the classic CSTelement is extended by a wrinkling model, a robust solver able to capture the resulting nonlinearities is implemented. This model is coupled to a commercial RANS solver by a bespoke coupling algorithm. This algorithm ensures the conservative transfer of forces and deformations while keeping the coupled simulation stable.
Next, to ensure applicability of the structural and flow simulation models as well as the coupling, they are verified for grid and time step dependency and validated against analytical or experimental data. As no experimental data was freely available on the particular case of downwind sails, wind tunnel tests were conducted to provide at least aggregate flow forces and flying shapes. Particularly the structural simulation and coupling were successfully verified and validated, the simulation of partially separated flow around highly curved surfaces like downwind sails exhibited a strong sensitivity to e.g. small changes of the angle of attack. Validation of the flow simulation was hampered by uncertainties in the experimental data.
Finally, the method is used to compare three sail designs on a hypothetical yacht based on the AC90rule. The impact of the sail design changes is clearly shown with small variations in sail (profile) depth resulting in very much different optimal angles of attack.
Improvements to the method could in particular be achieved by implicit or strong coupling of flow and structural simulation, this would yield timeaccurate information on the sails unsteady behaviour. Further, even more involved flow simulation methods, e.g. large or detached eddy simulation instead of turbulence modelling might improve the accuracy of the flow simulation.
Within this thesis a method is being developed to capture the interaction between the complex viscous flow around downwind sails and compute the structural answer to the resulting forces. First a structural model suitable for downwind sails is developed. This is coupled to a commercial solver for simulations of viscous flow. The individual parts (structural and flow simulation as well as coupling) and the entire method are verified and validated. Finally an application example is given. First, the structural model and coupling to the flow solver are developed. The particular challenge regarding the structural model is the requirement to compute the complex behaviour of downwind sails. By design these sails have negligible bending stiffness with the material being stiff in tension but without any meaningful compressive stiffness. To this end the classic CSTelement is extended by a wrinkling model, a robust solver able to capture the resulting nonlinearities is implemented. This model is coupled to a commercial RANS solver by a bespoke coupling algorithm. This algorithm ensures the conservative transfer of forces and deformations while keeping the coupled simulation stable.
Next, to ensure applicability of the structural and flow simulation models as well as the coupling, they are verified for grid and time step dependency and validated against analytical or experimental data. As no experimental data was freely available on the particular case of downwind sails, wind tunnel tests were conducted to provide at least aggregate flow forces and flying shapes. Particularly the structural simulation and coupling were successfully verified and validated, the simulation of partially separated flow around highly curved surfaces like downwind sails exhibited a strong sensitivity to e.g. small changes of the angle of attack. Validation of the flow simulation was hampered by uncertainties in the experimental data.
Finally, the method is used to compare three sail designs on a hypothetical yacht based on the AC90rule. The impact of the sail design changes is clearly shown with small variations in sail (profile) depth resulting in very much different optimal angles of attack.
Improvements to the method could in particular be achieved by implicit or strong coupling of flow and structural simulation, this would yield timeaccurate information on the sails unsteady behaviour. Further, even more involved flow simulation methods, e.g. large or detached eddy simulation instead of turbulence modelling might improve the accuracy of the flow simulation.
Original language  English 

Qualification  Doctor of Philosophy 
Awarding Institution 

Supervisors/Advisors 

Thesis sponsors  
Award date  12 Feb 2018 
DOIs  
Publication status  Published  2018 
Keywords
 CFD
 FSI
 Sailing