THE 20 th CHESAPEAKE SAILING YACHT SYMPOSIUM ANNAPOLIS S, MARYLAND, MARCH 2011 Photogram mmetry Based Flying Shapee Investigation of Downwin nd Sails in Wind Tunnel and at Full Scale on a Sailing Yacht Johannes Mausolf, Yacht Research Unit, University of Applied Sciences Kiel (UAS), Kiel / Germany Julien Deparday, ENSIETA, Brest / France Kai Graf, Institute of Naval Architecture, University of Applied Sciences Kiel (UAS), Kiel / Germany Hannes Renzsch, Delft University of Technology, Delft / The Nerlands Christoph Böhm, Delft University of Technology, Delft / The Nerlands 1 ABSTRACT This paper describes model and full scale meas- urements of flying shape of spinnakers s that have been developed for a 38 c/r yacht. The flying shape measurement principle was based on photogrammetry. Motivation behind this study was to investigate and compare flying shape as obtained in wind tunnel and at full scale. While flying shape measurements in wind tunnel are straight forward to do providing quite accurate results, respective measurements at full scale in natural, stochasticc environment turned out to be challenging. As a result of comparison it can be shown that flying f shapee at full scale and model agrees well for given trim settings of spinnaker, while trim settings for driving force optima where quite different. 2 Notation Abbreviation ns: DOF: TFWT: Degrees of Freedom Twist Flow Wind Tunnel NURBS: Non Uniform Rational B- Spline Symbols: AWA: AWS: TWA: TWS: AX Apparent Wind Angle Apparent Wind Speed True Wind Angle True Wind Speed Driving Force Area AY x, y x 0, X, Y, Z X 0, Y 0, Z 0 n SL y 0 SMW SF P E IM J MG[LMUT] HB z Side Force Area Imagee Coordinates Coordinates of perspective centree of image Objectt coordinates in space Coordinates of perspective centree in 3D space Exponent for boundary layer powerr law Spinnaker Luff and Leech Lengthh Spinnaker Mid Width Spinnaker Foot Length Main Luff Length Main Foot Length Jib Head to Deck distance Foresail triangle base length Main Girth Lengths Main Headboard Length Height above water plane 3 Introduction Wind tunnel testing of downwind sails is widely accepted as ann efficient tool for spinnaker design for contemporary sailingg yachts. It is conducted to generate aerodynamicc information for subse- quent sailing yacht velocity prediction, actually to generate set of aerodynamic coefficients, describing e flow forces sail generates. In addition, windd tunnel testing allows obtaining 1
ing, realizing heightt dependentt flow speed and direction,, a sailing yacht encounters on a downwind course. Maximum mast height of model is about 1.8 m. The model is mounted to a turntable, allowing arbitrary apparent wind angle of 0 to 180. A 6-DOF force balance is fixed to turntable. flying shape of spinnaker, i.e. shape s spin- see for example (Graf and Mueller, 2009). Flying shape measurements provide a means to analyze aerodynamic properties of a spinnakerr in naker will take under wind load and trimming, detail. It helps to understand conversion from design shape shape sailmaker actually defines to flying shape and thus helpss to design better sails. It also helps sailorss to understand how to trim sail to obtain maxi- mum performance. As ever model testing has severe drawbacks. In case of wind tunnel testing of sails a couple of rules of similitude are violated. Beside a far to small Reynolds number in model scale, usually neir ratio of fabric weight to wind pressure nor ratio of membrane stressess to wind pres- in sure in wind tunnel matches those t valuess reality. Since all see characteristics have an impact on flying shape of t sail, some doubts arouse, wher or not flying f shapee in wind tunnel matches one obtained in full scale. This paper reports about a study,, trying to en- lighten this problem. In cooperation with a sail- a maker (Holm Sailmaker GmbH,, Germany) series of symmetric spinnakers for a 38' c/r yacht has been investigated in wind tunnel. t For one showing best properties, a full scale replica has been build. Flying shape measurementss of this spinnaker in model and full scale have been conducted under various wind conditions. The measuremen nt principle is based on photogram- metry. Comparisons of model and full scale re- sults will be presented. 4 Flow Force and Flying Shape Meas- for uremts in Wind Tunnel The wind tunnel of YRU-Kiel and proceduress spinnaker tests including flying shape measure- 2009). YRU-Kiel's Twist-Flow Wind Tunnel is ments have been described by (Graf and Mueller, an open-jet wind tunnel, powered by two axial fans for a maximum wind speed of 10 m/s at measuring area, Fig. 4-1. Rectifiers, screens and twist vanes are used for proper flow condition- Fig. 4-1: TFWT of YRU-Kiel The model is equipped with stepperr motors con- trolled by PC-based virtual activators in order to trim sail. The following sheetss and haulers are available: Main-Sheet Boom vang Spinnaker-Sheet Spinnaker-Aft guyy Spinnaker pole vang Top lift Spinnaker-Barber r hauler 4.1 Photogrammetry The principle method of photogrammetry to ob- tain 3D-shape of a flying spinnaker in wind tunnel iss based on four components: a cou- 2
The cameras are connected to a PC using USB data bus. In addition y are equipped with a central trigger, allowing simultaneous trigger- sail vibrates under wind pressure. From an ing of any camera. This is quite important, since estimated frequency and amplitude of this un- time of 1/80 sec has been steady motion of sail a maximum exposure derived. The sail is equipped with a larger number of markers. These markers are distributed over sail surface such that a smooth surface can be generated fromm cloudd of marked points. Usu- ally 50 to 60 markers m are used. The software sys- of markers. The T patternn recognition algorithm behind this automatic detection needs so called coded targets as markers, having a diameter of tem used in this setup allows automatic detection approximatelyy 25 mm, see Fig. 4-3. These mark- ers are 12 bit coded allowing a oretical num- ber of 4096 different markers. Fig. F 4-3: Coded Targetss In addition too markers on sail, some markers are fixed to model in order to define a local coordinate frame.. Fig. 4-4 shows a sym- metric spinnaker of an IMS 600 model, equipped with coded targets. Note markers for co- ordinate framee on foredeck of model. Fig. 4-2: Camera arrangement in wind tunnel ple of images are taken from sail simultane- with a largerr set of markers at discrete pointss in ously by digital cameras. The sail is equipped sail. A chain of software tools are usedd to improve brightness and contrast off image, to automatically detect markers in i sail and finally kernel algorithm of photogram- points in images into 3D coordinates inn an metry to convert 2D coordinatess of individual absolute frame. Finally se points are loftedd to create spline curves, which in turn are loftedd to create a NURBS surface. 4.2 Setup in wind tunnel The method presented here uses four digital cameras Canon EOS 350D with a resolution of 8 million pixels and a zoom lens 17-85 mm focal distance. Fig. 4-2 shows arrangement of cameras in wind tunnel measurement section. Usually camera location has to be adaptedd to a range of apparent wind angles of model. However it can be chosen freely and actual location of camera has not to be known for proper shape detection. Fig. 4-4: Symmetric spinnaker equipped with 55 coded targets 3
This setup has successfully been used for a large number of different spinnakers, among m runners for very deep courses as well as flat asymmetric spinnakerss for quite low apparent wind angle. The measurement of flying shape is well l in- egrated into a standardd measurement run: a par- are trimmed for maximum driving force and n forces and moments generated by entire sail set are scanned. The photos are taken within force scanning period, whichh usually lasts approximately 10 sec. The cameras are equipped with 1 GB memory cards, allowing to take imag- es for an entire range of investigated apparent wind angles. For shape finding process Photo Modeler Pro (PMP) of EOS Systems Inc. / Canada is used. PMP is a MS-Windows based software with a ticular apparent wind angle is chosen, sails graphical user interface. It includes kernel photogrammetric algorithm which generates a 3D point cloud from identical (2D) points in a couple of images taken from sail from differ- ent views. PMP can take into account an arbi- trary number of different views / images. Two images are minimum, if camera positions are known a priori, three images are minimum, if camera position shall be calculated automati- cally by system. Any additional image in- four images of sail are sufficient. PMP generates a set of points in 3D space. This set can be exported as tabulated data or as IGES file, and in turn imported into a surface modeling system. For this purpose Rhinoceros 3D of creases accuracy. Tests show that for f our purpose McNeal Inc. is used. 5 Flying Shape Measurements at Full Scale For full scale flying shape measurement same measurement principle has been used. Four cameras take images simultaneously while spinnaker is set on sailing yacht. These cam- yacht at predefinedd positions,, moving eras are located on four tenders sailing around with same speed as yacht. Attempts to place one or two of o cameras on yacht itself have been abandoned after some tests. It was simply not possible to take cameraa shots from aboard showing a reasonable large number of markers on sail. Withh respect to camera posi- tunnel testingg was that all cameras obviously have to be located close to water surface. tions at full scale main difference to wind Transformation from e wind tunnel situation was used to define a general setup for f distri- 5-1. Fig. 5-1: Arrangement of tenders around bution of tenders around yacht, see Fig. yacht The PMP-method allowing taking images from arbitrary position as longg as at least three images are used for shape s finding provides some free- dom in actual position of a tender relative to yacht. In general g photographers were told to watch for a particular image contents to obtain a particular position. For full scale tests only conventional markers rar than coded targetss were used for sail surface, simply because coded targets are com- plicated to apply to spinnaker. As a conse- quence post-processing of images was a bit more time-consuming, since all markers must be 4
manually detected with help of PMP software. Fig. 5-2 shows yacht sailing at apparent wind angle of AWA=120, image taken from camera position 3 as to Fig. 5-1, Fig. 5-3 showing same from camera position 2 and 4. Note mark- ers at guardrail for reference coordinate sys- tem. The main challenge of field measuremen nts has been to cope with stochasticc behavior of wind conditions in natural environment. While in wind tunnel constant laboratory conditions prevail for some time and accurate flow force measurementss allow to find an optimum, in full scale field tests amount of time to trim spinnaker for optimum driving force is limited, wind conditions change frequently and finding of what is regarded to be an optimumm trim has a strong human factor. Measurementss took placee in inner part of Kieler Förde where w flat water could be expected. The followingg procedure has been applied: Move to starting point of a measure- prevail. Set spinnaker, find an apparent wind an- gle that can kept constant for a couple of ment run, where constant wind conditions minutes. Trim e spinnaker for optimum boat speed. Call tenders to obtain ir intended position around yacht. Trigger a simultaneous shot of all camer- conditions, boat speed and heel from nau- tical sensors and define a qualitative as- as with an acoustic signal, read wind sessment of this measurement ("good", "sufficient", "instationary", ) Repeatt triggering up to four times. Find a new location and a new apparent wind angle a and repeat entire procedure. Fig. 5-2: Spinnaker with non-codedd markers, camera pos. 3 Following thiss procedure, a total off 80 test runs were done successfully, which were distributed over a range of apparent wind angle 80 <AWA<160 and 3m/ m/s<aws<10m/s accord- ing to Fig. 5-4. Fig. 5-3: Images from camera pos. 2 and 4 5
AWS [m/s] 12 10 8 6 4 ok poor good Fig. 7-1 shows driving force area and side force area AX and AY over apparent wind angle AWA. Here AX is defined as: 1 2 2 70 90 110 130 150 170 AWA [ ] Fig. 5-4: Test conditions at full scale 6 Spinnaker Models Two series of 8 symmetric molds have been developed for this investigation. The first series focuses on total size of spinnaker for given leech/luff and foot length, leaving spinnaker maximum width as variable parameter. Shape A4-mod emerged from A4 by manual manipulation after visual inspection (fixing wrinkling using scissor and tape). The second series consists of only small variations of A4-mod. A generic mold has been developed for A4-mod in order to make use of it in full scale within a sail lofting program (A5). Variations of A5 differ only very marginally in topmost part of spinnaker. When designing se alternatives, target was to have a very flat, as broad as possible head without tendency of spinnaker to collapse early. Serie 1 Serie 2 Name SL SF SMW Area Holm Area Calc. SMW/SF [mm] [mm] [mm] [m²] [m²] [ ] A1 1430 720 700 0.88 0.948 97% A2 1426 808 963 1.15 1.239 119% A3 1430 808 808 1.01 1.086 100% A4 1430 808 880 1.08 1.159 109% A4 mod 1430 808 808 1.01 1.086 100% A5 1430 808 880 1.06 1.159 109% A6 1430 808 820 1.098 101% A7 1430 808 820 1.098 101% A8 1430 808 820 1.098 101% Fig. 6-1: Model Spinnaker Dimensions 7 Wind Tunnel Flow Force Results The spinnaker have been tested in wind velocity of 5 m/s at mast top and a total twist angle of approximately 15. The spinnakers were trimmed for maximum driving force. where FX is measured flow force in yacht longitudinal axis. Ax [m²]; Ay [m²] 2.00 1.50 1.00 0.50 0.00 0.50 A1 A3 A4 A4 mod A5 1.00 60 80 100 AWA 120[ ] 140 160 180 Fig. 7-1: Wind Tunnel Force Measurement Results Series 1 The result clearly shows that neir maximum sized spinnaker gives best performance on deep courses nor most slender spinnaker on hot course. A4-mod did show best overall performance. It provided most driving force on deep courses, while being close to best spinnaker with respect to driving force on hotter courses, albeit at significantly lower side forces than spinnaker A5, providing maximum driving force at hot courses. Variations of A4-mod have been tested in a second series, result shown in Fig. 7-2. Here differences are very small as expected. The small variations of topmost part of spinnaker showed some impact on dynamic property of shape, which has been observed visually only. Finally shape A7 has been identified as top-scorer in test field, providing best compromise on flow forces and dynamic behavior. This spinnaker n has been produced in 6
full scale (scale factor 11) to be used for scale flying shape measurements. full full re-trim sheet length. of spinnaker pole and Ax [m²]; Ay [m²] 2.00 1.50 1.00 0.50 0.00 0.50 Fig. 8-1 to Fig. 8-11shows flying shape of model and full scale (scaled to model size) for range of AWA=80, A 100, 120 and 160 and all three modes of comparison. Fig. 8-12 to Fig. 8-16 shows a contour plot of Mode I comparison. 1.00 A4 mod A6 A7 A8 1.50 60 80 100 120 140 AWA [ ] Fig. 7-2: Wind Tunnel Force Measurement Re- sults Series 2 8 Flying Shape in Wind Tunnel and at Full Scale Flying shape measurements in model and full scale have been carried out with spinnaker shape A7 at apparent wind angles of AWA=80, 100, 120, 140 and 160. The comparisonn is shown in following diagrams. 160 180 In this study wind tunnel flying shape measure- ments have been done after completion of full scale flying shape measurements, in order to maintain comparable conditions. Here it turned out, that re are actually three modes of com- parison: Fig. 8-1: Mode I comparison at AWA80 Mode I comparison: In wind tunnel spinnaker pole height and angle with respect to yacht center line is sett to respec- tive values at full scale. The same holds for spinnaker sheet length and sheet lead position. Spinnaker clew is free to move under sheet constraint. Mode II comparison: Spinnaker pole is sett as under Model I, however sheet- length is varied to search for maxi- mum driving force. Mode III comparison: Starting from trim as in Mode I model spinnaker has undergone a full optimization, i.e. a 7
Fig. 8-2: Mode II comparison at AWA=80 Fig. 8-4: Mode I comparison at AWA=100 Fig. 8-3: Mode III comparison at AWA=80 Fig. 8-5: Mode II comparison at AWA=100 8
Fig. 8-6: Mode III comparison at AWA=100 Fig. 8-8: Model II comparison AWA=120 Fig. 8-7: Mode I comparison AWA=120 Fig. 8-9: Mode III comparison AWA=120 9
Fig. 8-12: Mode I, AWA= =80 Fig. 8-10: Mode I comparison AWA=160 Fig. 8-13: Mode I, AWA=100 Fig. 8-11: Model III comparison AWA=160 10
Fig. 8-16: Modee I, AWA=160 Fig. 8-14: Mode I, AWA=120 Fig. 8-15: Mode I, AWA=140 Analysis of diagrams d for Mode I comparison gives a quite satisfying s agreement between mod- el and full scale flying shape. Differences are within a rangee of 1 to 2% % of leech length for most parts of surface, with maxima reaching 4-5%, which mainly m can n be found at foot of sail. Considering e challenging test condi- result. It rectifies twist chosen in wind tunnel to properly fit one found in full scale. It also allowss concluding that violation of tion one encounters at full scale this is a good rules of similitude as discussed in chapter 1 can be accepted att least to a high degree.. On orr hand comparison results of Mode II and III I raise some doubt in validity of optimization procedure in wind tunnel and at full scale. As a general trend it has been shown that it was w possible to ease sheet much more in wind w tunnell as in full scale and to haul aft guy tighter, all this with conse- quence of higher driving g forces compared to trim that has been regarded to be optimum while sailing full f scale. There are two possible reasons for this: a) due too low Reynolds number pressure field close to luff of sail develops 11
in a way that collapsing of luff is delayed to much higher local angles of incidence. In this case optimization procedure as used in wind tunnel remains questionable. b) It has to be considered that trimming of spinnaker on sailing yacht ( maximization for driving force in full scale) was done under high time pressure in an in-stationary, somewhat stochastic environment. The trimmers on board have been reasonable experienced sailors but no professional spinnaker trimmers. Consequently it may be possible that trim for maximum driving force has not been achieved in full scale. 9 Conclusion This paper describes a study about comparison of flying shape as found in wind tunnel and in full scale. The measurement principle was based on photogrammetry. As a result of study it was found that good agreements of model and full scale flying shape can be achieved for comparable spinnaker trim conditions. However re are some hints that raise doubt about comparability of trim that has been regarded to be an optimum one in full scale and trim that has been thoroughly optimized with use of instrumentation and constant laboratory conditions in wind tunnel. Future research on this topic should try to replicate a trim at full scale that has been found to be an optimum in wind tunnel. Due to stochastic nature of field tests this can only be done by mass-data investigation and in a sailing environment with very stationary wind and wear conditions. 10 References Graf, K. and Mueller, O. (2009): "Photogrammetric Investigation of Flying Shape of Spinnakers in a Twisted Flow Wind Tunnel", Proc. 19 th Chesapeake Sailing Yacht Symposium, Annapolis / USA, 2009 12