Shuttle Tanker Design Problems Solved by CFD-Code DAWSON

SNAME Transactions, Vol. 102, 1994, pp. 71-85 Shuttle Tanker Design Problems Solved by CFD-Code DAWSON H. H. Valkhof, Visitor, MARIN, Wageningen, The Netherlands, and E. Minguito, Member, Astilleros Espafioles, Madrid ABSTRACT This paper will describe the procedure followed to solve certain design problems arisen during the development of two 126,500 DWT shuttle tankers to be built at ASTILLEROS ESPAIqOLES for KNUTSEN O.A.S. Shipping. These problems became apparent while testing the slightly modified version of an existing tanker hull form at MARIN. The original hull form was also extensively tested in said Institute. The modification comprised a lengthening of the parallel midbody, a shortening of the bulbous bow length, and a movement of the rudder location aftwards, while further one additional bow thruster opening (3 instead of 2), a bottom cargo system opening (Submerged Turret Loading), and one additional stern thruster opening (2 instead of 1) was fitted on the existing ship model. The results of both the resistance and propulsion tests carried out with the modified hull form were quite disappointing: the ship required about 10% more power at the contractual speed than expected. An additional series of model tests was therefore carried out to more accurately determine all the separate effects of the openings fitted to the model. From these tests it became clear that the 3 bow thruster tunnel openings were mainly responsible for the unexpected significant resistance increase. This problem was efficiently solved by using mathematical tools. In fact, a potential flow calculation carried out with MARIN's DAWSON-code for the tested hull form showed a very low pressure and a steep pressure gradient at the position of the forward bow thruster opening. The direction of the flow at the same position was nearly vertical. Two alternative forebody hull lines were then designed for which potential flow calculations were carried out too. The results of these calculations were so promising, showing a softer depression with improved pressure gradients and a more longitudinally directed flow, that it was decided to modify the ship model. The modification was in fact a compromise between the two designed ones, due to some constructive requirements. The results of the subsequent model tests were in line with the high expectations based on the potential flow calculations and the gain in ship's speed was even larger than expected. The maior conclusion drawn is that without this CFD code these problems would not have been solved so easily and so quickly. 1. INTRODUCTION In September 1992, ASTILLEROS ESPAIqOLES signed a contract with KNUTSEN O.A.S. Shipping for the building of two 126,500 DWT shuttle tankers in its shipyard of Puerto Real, with the option of a third one. Considering the very good propulsive performance of a series of two tankers and three shuttle tankers built in the previous four years by ASTILLEROS ESPAiqOLES, it was decided to base the hydrodynamic design on the hull form of said series. The story therefore started five years earlier: by the end of 1987, when ASTILLEROS ESPAlr, IOLES asked MARIN to cooperate in the development of the lines for a series of two 112,500 DWT tankers ordered by KNUTSEN O.A.S. Shipping (Sestao shipyard newbuilding 275, "TOVE KNUTSEN", and 278, "DICTO KNUTSEN"). The lines were designed in a traditional way, which means that no Computational Fluid Dynamical tools were used and hence the design was based on the knowledge of and experiences with more or less similar ship types available at ASTILLEROS ESPAJqOLES and MARIN (Fig. 1). The resistance, propulsion and wake measurement tests showed quite favorable results, when compared to similar hull forms available in MARIN's data base. Also from the full scale trial results only marginal differences were found with the predictions made by MARIN based on the model test results. Based on the very good propulsive performance of the hull form designed for the previous ships, ASTILLEROS ESPAIqOLES decided to apply the same lines for the building of three new 125,000 DWT shuttle tankers ordered also by KNUTSEN O.A.S. Shipping in 1989. These tankers to be built at the Sestao shipyard were modified however by lengthening the parallel midbody by 18.5 m, shortening the bulbous length, and incorporating one additional bow thruster tunnel and one additional stern thruster tunnel, (Fig. 1). The modified ship model was also tested at MARIN in order to determine the influence of such modifications on the propulsive performance: an average increase of power at constant speed of about 13.5% was found, being in line with the estimations made prior to the model tests. Also in this case the full scale trial results of the three new ships: "MARIE KNUTSEN" (newbuilding 285), "TORDIS KNUTSEN" (newbuilding 288) and "VIGDIS KNUTSEN" (newbuilding 289), were found to be well in line with the predictions presented in the corresponding MARIN report (Koops, Valkhof and Mulder, 1991). 71

BODY PLAN. STEM AND STERN PROFILES AND SECTIONAL AREA CURVE OF SHIP MODEL 2. SHIP'S OPERATIONAL DEMANDS, ADVISORY STUDY AND MODEL TEST RESULTS s~l O ^P Stll 0 A P ORIGINAL SHIP MODEL FINAL LENGTHENED SHIP MODEL Star 20. F P staq o. ^ P sl.~ 2o ~ g P Fig. 1. Hull form of original ship model. Hull form of lengthened ship model. Hull form of final lengthened ship model. Thus arrives 1992, when ASTILLEROS ESPAIqOLES got a new order to build two 126,500 DWT shuttle tankers for KNUTSEN O.A.S. Shipping, and it was decided to use the previous hull form as a good starting point. Again the parallel midbody was lengthened by 5 m, the bulbous bow length further reduced, the rudder position moved backward, the transom was modified, and one new bow thruster, one new stern thruster and a bottom opening to locate the Submerged Cargo System were incorporated (Figs. 1 and 2). The second section of this paper will describe the operational demands of the new ships, the results of the estimations carried out by MARIN to determine, based on statistical values, the effect of the modifications of the hull and the openings fitted on the ship model, while further in this section the results of the first series of model tests are presented. The third section will describe the procedure followed and the techniques used to solve the problems arisen. In this section also a brief history of potential flow calculations, in particular the related MARIN code DAWSON, will be presented. Besides this general description the results of the calculations carried out for this specific shuttle tanker project will be presented and analyzed. In the fourth section of this paper the results of the model tests performed with the ship model optimized via DAWSON calculations are shown while finally in section 5 the conclusions are drawn. 2.1. Ship's operational demands The vessel is a 126,500 DWT double hull and double bottom crude oil tanker, provided with segregated ballast tanks and diesel-electric propulsion together with a fixed-pitch propeller. The ship has been designed to be operated at Heydrum Field, in the Norwegian Sector of the North Sea, at the very north, near the Arctic Circle. Crude oil loading operations can be performed from an offshore installation, either through a Bow Loading System (BLS), or through the bottom by means of the new Submerged Turret Loading system (STL), or through manifolds, as any other conventional crude oil tanker (Fig. 2). The cargo space between Pump Room and Fore Peak is divided into 18 cargo tanks, with 140,000 m 3 capacity together, and 12 segregated ballast tanks, with 60,000 m 3 capacity together, by means of two longitudinal bulkheads, the inner shell, five transverse bulkheads, and two slop tanks behind the cargo tanks. ASTILLEROS ESPAlqOLES is at present in the lead of the world shuttle tankers market, since it has built and is building one fourth out of all ships now prepared to work in significant waves up to 5 meters height. The ship thus is fitted with three bow thrusters and two stern thrusters, each of them of 1,700 kw, a fixed-pitch propeller of 8.00 m diameter and a Becket type rudder, in order to operate the ship in such waves, and to perform very extreme loading operations within the required safety margins, by means of a very effective and sophisticated Dynamic Positioning system, resulting in the following capabilities: a) During connection stage to the loading buoy, the vessel will be able to keep her position in significant waves up to 6 meters, with winds blowing at 20 ms, waves coming from the same direction and making an angle of 10 deg with the bow. b) In conditions of a storm with a return period of 100 years, significant waves up to 15.5 m, wind blowing at 37 ms, currents generated by the wind of 0.7 ms and a maximum separation angle of 20 deg, the five thrusters will be able to reduce the second order movements of the vessel, so that stresses supported by the anchoring system be within the maximum auowed. Assuming a thruster output ratio of 0.15 knkw, the ship is so provided with a maximum lateral thrust of 1,275 kn. As far as main propeller and all thrusters are concerned, a total power of 22,000 kw is required to survive a storm with a return period of 100 years, considering a minimum draught of 12m. The ship is also fitted with a propulsion electric motor of 19,000 kw at 98 RPM. Main generating plant is composed of four engine-generator groups. Each generating plant comprises a Sulzer diesel 9ZAL4OS, of 6,598 kw at 514 RPM, which drives an ABB generator of 6,275 kw, 6,600 Volts and 60 Hz by means of a flexible coupling. A more detailed description of the installations and special features of this vessel can be found in (Galfin, 1993). 72 Shuttle Tanker Design Problems

GENERAL ARRANGEMENT SHUTTLE TANKER ~, "--~...! [4-'-...-'~; ".-. ;~,~-.-..-, ".-..- ~'.-. ~." ~\, ~-';-~ }:"!.~:4['~="~'~7 ~:"r-r-r-r:r."~nff~,-rt-'--'--sq~.'~t~-:~r:~'-r-r:-r~t:'7-@'.~..~':-:':-n)')l~y, I PROFILE. ~...~..._ ~...., ~.......... ~@) PLATFORM DECK UPPER DECK FORECASTLE DECK OOUBLEBOTTOM Fig. 2. General arrangement. 2.2. Advisory study Prior to the model tests with this modified and adjusted hull form, a study based on both the test results of the original and the first modified ship model was carried out to predict the separate effects of the hull form changes and the openings fitted. In the table below the differences between the main particulars of the original, first modified and final modified hull form are presented: Hull form Lpp Length Stern Bow STL bulb thruster thruster (m) (m) Original 233.0 5.0 1 (i) First modification 251.5 3.5 1 2 (11) Second modification 257.0 1.8 2 3 1 (11i) The model tests carried out with the original hull form showed a detrimental effect of the bow thruster tunnel on the propulsive performance of about 2 per cent. As mentioned before, the first modified and enlarged ship model fitted with 3 thruster tunnels required an average of about 13.5 per cent more power at constant speed than the original one: 4 per cent due to the two new thruster openings and 9.5 per cent due to the lengthening of the parallel midbody and the reduction of the bow length. Based on the above information and taking into account other statistical data available at MARIN, the following estimation was made to evaluate the increase of power corresponding to the second and last modified ship model: 5 m lengthening of ship + 1% Shortening of bulbous bow + 1% STL opening + 4% 1 additional bow thruster opening + 2% 1 additional stern thruster opening + 1% Total estimated power increase + 9% 2.3. Model test results (first series of tests) In order to evaluate the ship's performance, a model test program comprising resistance, propulsion and wake measurements in full load and ballast (only resistance) conditions, was carried out at MARIN facilities. A short resistance test at the full load condition with the STL opening closed Shuttle Tanker Design Problems 73

was also carried out to determine its contribution to the ship's resistance. Below a comparison of the effective power in percentages is presented between the results obtained with the STL opening closed and open: Speed (kn) Bow thruster closed Stern thruster closed STL closed 14.0 100.0% 15.0 100.0% 16.0 100.0% Speed STL open STLclosed (kn) (%) (%) Bow thruster closed Stern thruster closed STL open 102.5% 102.5% 102.1% 14.0 100 98.0 15.0 100 97.9 Bow thruster closed Stern thruster open STL closed 104.0% 104.2% 102.3% 16.0 100 98.3 From these results it can be concluded that the effect on the resistance of the STL opening, expected to be approximately 4%, is considerably lower: about 2%. Given these favorable results said opening was accepted without further modification. The propulsion and resistance test results with all tunnels open were however quite disappointing: an average increase of about 18% in effective power (see the following table) at the design draught condition (15 m) was found when comparing the results obtained with the previous ship model; that is, 9% more power than expected based on both the previous test results and the MARIN statistics. This gap between the expected and the actual results was even 2% larger if the very favorable STL opening performance is taken into account. Speed Ship model ~ Sh~modelgI (kn) (%) (%) 13.0 100 118.2 14.0 100 118.3 15.0 100 118.0 In terms of speed, the previously mentioned gap in resistance means a decrease of 0.4 knots at the contractual power: a trial speed of 15 knots was predicted for the ship based on the model test results, while based on MARIN's estimation a speed of about 15.4 knots was achievable. In order to solve this problem, it was felt to be necessary to determine the separate effects on the resistance of all the openings fitted on the ship model more accurately. For this reason, an additional series of short resistance tests was carried out at 15 m draught by closing alternatively the bow thruster tunnel openings and the stern ones. In the following table, a comparison of the effective power obtained in these resistance tests is made. The results of the bare hull configuration are valued as 100%. Bow thruster open Stern thruster open STL closed 122.2% 122.1% 120.3% This comparison clearly shows that the bow thruster tunnel openings, in particular the most forward one, since the other two are located at the same position as in the previous ship model, are for the most important part (approximately 17 to 18%) responsible for the unexpected dramatic increase in resistance. Though a part of the flow around and in the bow thruster tunnel is scale dependent, for which the results have been improved in the extrapolation by some 1 to 1.5 per cent, the major part of the resistance increase is due to pressure drag, being largely scale independent. The absence of experimental results for the previous ship model, at the same ballast draughts as tested now, made it impossible to assess the resistance penalty in this load condition. However, it could be anticipated that the problem detected at 15 m draught is also present in ballast condition (see next section). At that moment the most important question arose: How to solve the problem?. The contractual condition could not be attained with this hull form tested and therefore a solution had to be found. Since MAR1N, like many other Institutes these days, possessed its own developed potential flow codes (linearized DAWSON (Dawson, 1977) and nonlinearized RAPID (Raven, 1988)), which had already been used at that moment by ASTILLEROS ESPAlqOLES in close cooperation with MARIN as an standard tool for optimizing the hull lines of all its projects (Minguito, 1992), it was decided to use the MARIN code DAWSON to improve the flow characteristics around the bow thruster tunnel openings, by calculating a few modifications of the forebody lines suggested by both MARIN and ASTILLEROS ESPAqOLES. 3. POTENTIAL FLOW CALCULATIONS The resistance of a ship advancing steadily in still water is usually decomposed into a viscous and a wave resistance component. According to Froude's principle for conducting model tests, these components may in first approximation be assumed to be independent of each other and to be governed by the Reynolds number and the Froude number, respectively. In the MARIN formulation of the problem of calculating the wave making and wave resistance of a ship, it 74 Shuttle Tanker Design Problems

is supposed that the flow is inviscid. Furthermore, the presence of wave breaking has not been taken into account. It follows that the irrotationality of the incoming flow is preserved, so a potential flow may be assumed: the velocity in the flow field is the gradient of a velocity potential. The computer code then has to solve the Laplace equation for the velocity potential in the fluid domain subject to boundary conditions on both the hull and the free surface. The free surface conditions must be satisfied at the water surface, but the location of this is unknown and is a part of the solution. The exact mathematical model constitutes therefore a nonlinear free surface problem, which is not easy to solve. The problem can be simplified by linearizing the free surface conditions. Simple as this may seem, in the pre-computer age even this problem was too complicated to solve. Therefore further simplifications were made. A whole class of linearized theories were proposed, such as: thin ship theory, fiat ship theory, slender theory and other variants. Though the thin ship theory was the most successful of these methods, it appeared to be too restrictive to be practically useful, as only accurate results could be obtained for unrealistic high length-beam ratios. In the late seventies another type of linearization was proposed. Instead of linearizing with respect to uniform flow, it proved better to choose a base flow that at least goes around the hull. This is true for the "zero Froude number flow", the wave-free flow around the hull; this is equivalent to the flow about a "double body" made up of the hull and its mirror image in the waterplane, advancing through an unbounded flow domain. Since the base flow corresponds with a zero Froude number, the linearization will be adequate for small Froude numbers, and this approach is called "slow ship theory". In practice however, it has appeared to work well for Froude numbers occurring in normal shipping practice. This base flow can be calculated by e.g. the method of Hess and Smith, by putting Rankine source panels on the double body surface. The resulting system of equations is solved for the source strengths, from which the velocity and pressure field without wave effect are easily deduced. The free surface conditions can then be linearized with respect to this flow field and the associated "double body wave elevation". The solution consists of two steps: first the calculation of the double-body flow, then the solution of the free-surface problem. The essence of the method of DAWSON is to use simple Rankine sources also in the second step. Hence the free surface conditions is not inherently satisfied, and to impose it, a part of the free surface surrounding the hull must now be covered with source panels too. The code thus ends up with a large system of equations for the unknown hull and surface source strengths. After solving this, the velocity and pressure field are easily found for the desired Froude number. The wave resistance and other forces on the hull can then be deduced by pressure integration or otherwise. The fact that many DAWSON like methods have been developed does not mean that these are all equivalent. The predicted wave resistance is basically rather sensitive to numerical inaccuracies. Experience with the method is needed in order to get reliable predictions. Important is frequent application in commercial projects, such that experience is gained on the numerical accuracy and the correlation and interpretation of the results. The code DAWSON has shown to be an outstanding tool for reducing the resistance of a vessel by either local or major hull form modifications in a very short lime (Berg van den, Raven and Valkhof, 1990 and Ligtelijn, Raven and Valkhof, 1991). In 1993, more than 100 calculations were performed at MARIN in the day-to-day ship design work for the industry. Much experience makes it possible to pre-optimize hull forms efficiently before model tests are conducted. This leads to more economic designs, more economic model test programs and a significantly shorter design period. In the present project the program was used for a different purpose when compared to the normal routine during a hull form optimization process. The program yields also the flow characteristics, i.e. the pressure field, the pressure gradient and the flow direction is calculated by the program. This information was used to optimize the flow around the bow thruster openings trying to reduce as much as possible the negative effects observed. In order to know more on what was happening, MARIN carried out a potential flow calculation with the tested hull. The results (Fig. 3) showed very low pressures with a steep gradient around the most forward bow thruster tunnel. SHUTTLE TANKER FIRST CONFIGURATION JL_2 Fn = 0.1569 DRAUGHT = 15.00 m Cp = 1 - (Vs) O95 0 j ij ~85 195 19 18_~ 18.~ o4 o.2!-^6: i : i i :rl,, :,'~) Y ~:.. "... :---.... :,: ol 02 ]4 V" ' ", ; "l I,",,' '\ il ~... 19 195 20' Fig. 3. DAWSON pressure distribution of shuttle tanker first configuration, showing the three bow thruster tunnels. Shuttle Tanker Design Problems 75

Therefore it was decided to especially modify this region in the forebody in an attempt to reduce the low pressure area around the bow thruster openings and furthermore to direct the flow more longitudinally compared to the original flow, which was directed more downward. There were, however, a few general arrangement constralnts which made it difficult indeed to find a solution: it was not possible to move the bow thruster tunnels location aftwards (the longitudinal flow component becomes more predominant as the flow progresses along the hull), as well as to increase the bulb length due to the location of the Bow Loading System on the castle deck, which is the other usual way to make the flow in the bow thruster tunnels region more longitudinal. What to do? After discussing different alternative solutions, it was finally agreed to change the small ballast bulb shape by increasing its volume and moving up its center of gravity (Fig. 4). The results of the potential flow calculation carried out for this modified forebody (Fig. 5) showed a reduction of the low pressure of the tested hull form from Cp (= 1 - (VVs) 2) -1.10 to -0.75. Further the results indicated too that the pressure gradient had been smoothed, while the flow was directed more longitudinally. BODY PLAN 19.5 19 18.5 18 Fn = 0.1569.i FIRST BULB CONFIGURATION DRAUGHT = 15.00 m i,i 195 ol :11 : :Y: ili!i " 2 G2 ~4 Oo '! III I {11..., Ill '" \ X X Cp = 1. (~ss) 2 19 185 18 "',... ~04 " } I, 5 185 19 195 2O Fig. 5. DAWSON pressure distribution of first bulb modification, showing the three bow thruster tunnels. STEM PROFILE 18.5 19 19.5 Fig. 4. First bulb modificatior~ j 20 Based on these promising results, a second alternative forebody hull form was designed (Fig. 6) by slightly modifying again the bulb shape in order to make the previous positive effects more favorable. The results of the potential flow calculation carded out immediately after (Fig. 7) showed a further reduction of the low pressure from Cp -0.75 to -0.7, and an additional smoothing of the pressure gradient, while the flow was even more longitudinal. Considering these results, and given the construction requirements concerning the most forward bow thruster tunnel, which made it impossible to adopt the last alternative, a compromise between the two described solutions was selected as final forebody hull lines (Fig. 8). To further improve the flow behavior around the bow thruster tunnels a fairing was proposed (Fig. 9). Based on the above mentioned final forebody lines, including the fairing around the bow thruster tunnel openings, the original ship model was modified and a new model test program was carried out. 76 Shuttle Tanker Design Problems

SECOND BULB CONFIGURATION BODY PLAN Fn = 0.1569 DRAUGHT = 15 00 m Cp = 1 - (~-s)2 I 195 t ) 18 5 18 195, A 185, to :",,'c 3" "': 185 195 1 I J 2O J,,yj J _J 1 -o~ o O2 a, l Fig. 6. Second bulb modification. ~85 A 195 BODY PLAN. Fig. 7. DAWSON pressure distribution of second bulb modification, showing the three bow thruster tunnels. I) ~,~ lo.s] THREE BOW THRUSTER TUNNELS I AND ONE TURRET CONE I \ STEM PROFILE 7 ~ ~_~ J l 1 I 18 5 19 19.5 J Stat 20 = F P 18 185._-L '-J -j 19 19.5 20 Fig. 8. Final bulb modification Fig. 9. Bow thruster tunnel openings fairing. Shuttle Tanker Design Problems 77

4. SECOND SERIES OF MODEL TESTS With this new ship model resistance and propulsion tests were carried out in both ballast (8.59.6 m) and full load (15 m) conditions. Below a comparison in effective power in presented between the results obtained with the originally tested ship model and the modified one, while for completeness sake the results of the ship with a length of 251.5 m are presented too but only in the full load condition comparison, since her ballast draughts are quite different: Speed (kn) Original ship model 3 Bow thrusters 2 stern thrusters 1 STL Modified ship model 3 Bow thrusters 2 Stern thrusters 1 STL Speed (kn) Ship model 251.5 m 2 Bow thrusters 1 stern thruster No STL Original ship model 3 Bow thrusters 2 Stern thrusters -1STL Modified ship model 3 Bow thrusters 2 Stern thrusters 1 STL Ballast condition (8.59.6 m) 15.0 16.0 100.0% 100.0% 92.9% 94.6% Full load condition (15 m) 17.0 100.0% 94.4% 14.0 15.0 16.0 100.0% 100.0% 100.0% 121.2% 120.9% 119.1% 106.7% 107.8% 108.6% From this comparison it can be concluded that the results for the modified hull form have been improved by 10 to 15 per cent in the full load condition, and 5 to 8 per cent in the ballast one, when compared to the results of the originally tested ship model. The differences between ballast and full load are due to the more longitudinal character of the flow in the original hull form ballast condition when compared to the full load one, as usually occurs for this kind of ships. It can be concluded also that with regard to the previous ships, the hull form modification including the two additional tunnels, the STL opening, the lengthening of the parallel midbody and the modification of the bulb in length and shape have increased the resistance by some 7 to 8 per cent, which is even somewhat better than the earlier expectations (see section 2). In terms of required power, the differences are similar in the full load condition: between 9 and 13%, as shown in the next table and in Fig. 10. Concerning the ballast condition, it can be assumed that the differences between the original tested hull form and the modified one remain the same as in resistance (5 to 8 per cent). Speed (kn) Original ship model 3 Bow thrusters 2 Stern thrusters 1 STL Modified ship model 3 Bow thrusters 2 Stern thrusters 1 STL Full load condition (15 m) 14.0 100.0% 87.3% 15.0 100.0% 88.2% 16.0 100.0% 91.4% As a consequence of the improvement of the propulsive performance, the ship's speed at 15 m draught with the propeller absorbing 17,100 kw in trial conditions (sea and wind not exceeding NB2) increases from 15.06 kn (original hull form) to 15.48 kn (modified hull form); that is, about four-tenth of a knot, which is a very important improvement achieved by using solely mathematical tools: model tests were carried out only to isolate the resistance increase cause (bow thruster tunnel openings), and to validate the predictions made by the MARIN code DAWSON, which accuracy is good in a qualitative sense (relative figures) but not in a quantitative sense (absolute values). It has to be emphasized that the process followed to solve the problem of the dramatic increase of resistance and therefore required power was carried out in only a couple of weeks, including the time needed to run the DAWSON code program. If such a mathematical tool had not been available at that moment, and so, this research study had to be made by means of model tests, the previous solution had very likely never been found, since the type of required information on the flow particulars is hard to obtain from experimental results. In any case, the process to solve the above described problems by applying physical tools would have consumed a lot of time: months instead of weeks. Before ending this section, it also can be stated that compared with the statistical data of MARIN, the results of the resistance and propulsion tests at the draught of 15 m for the final hull configuration can be qualified as good to even very good when taking into account all the openings fitted to the ship model. It can further be noticed that the extrapolated ship speed is better than the speed estimation based on the statistical data of MARIN and presented in the section Advisory work of this paper. 78 Shuttle Tanker Design Problems

SPEED-EFFECTIVE POWER PREDICTIONS (*) Design draught condition (15 m.) 16.000 14.000 13.000 PE (KW) Prevloue Ihlp --~ Original Iloe - " ~ Modified Iinell 20.000 SPEED-POWER PREDICTIONS (*) Design draught condition (15 rn.) POWER (KW) +Or o o" 'n'" 1 ± Meal,I,.0,,--. j ~ 12.000 18.000 11.000 10.000 10.000 9.000 14.000 8.000 12.000 7.000 6 000 10.000 6.000 13 r t 13,6 14 14 6 VS (knots) 8,000 13 I 13.6 14 14 6 15 16.6 16 VS (knots) I (') Balled on model tellt resultl (,) Based on model 1e 1 telullll Fig. I0. Speed power predictions at 15 m draught. 5. CONCLUSIONS a) All kind of openings to be fitted on the hull below the waterline, like thruster tunnels, cargo systems,... etc, have to be located in accordance with the surrounding flow particulars, avoiding regions with low pressures and steep gradients. In case that one of such openings is fitted on said regions, a dramatic increase of resistance, up to 15%, can be expected, as described in the present paper. b) An effective way to trace the source of the above-mentioned increases of resistance is the use of modem mathematical tools like MARIN's DAWSON-code or similar. By studying the flow particulars around the openings, decisions of the modifications needed to solve the problem, can be readily taken. c) Usually insufficient information concerning the flow particulars is supplied by the traditional model tests, which have to be considered as inadequate tools to solve problems as described in this paper. d) The required power increase of about 10% found in the model tests carried out with the original ship model when compared with expectations based on the wider experimental data available on such hull forms, has been eliminated and even improved by modifying the bulb shape and moving up its volume, in order to induce a more longitudinal flow in the bow thruster tunnels region. Such an important improvement has been achieved by only using mathematical tools. e) The above mentioned very unusual problem was solved in only a couple of weeks, including the time needed to prepare the lines to run the DAWSON code. In case that such a problem had to be solved by model tests, the time consumed would have been much longer: months instead of weeks. REFERENCES BERG van den, W., RAVEN, H.C. and VALKHOF, H.H. 1991 Free-Surface Potential Flow Calculations for Merchant Vessels Proceedings International Symposium on CFD and CAD in Ship Design, Wageningen. DAWSON, C.W. 1977 A Practical Computer Method for Solving Ship-Wave Problems Proceedings Second International Conference on Numerical Ship Hydrodynamics Berkeley, U.S.A. GAL,~N, G. 1993 The first modem diesel-electric tanker. KOOPS, A., VALKHOF, H.H. and MULDER, W. 1991 MARIN report No. 08585-6-VT: Model Tests for a 125,000 DWT Tanker. LIGTELIJN, J.Th., RAVEN, H.C. and VALKHOF, H.H. 1991 Ship Design Today: Practical Applications of Computational Fluid Dynamics Proceedings HADMAR '91 Symposium, Varna, Bulgaria. Shuttle Tanker Design Problems 79

MINGUITO, E. 1992 Use of CFD Tools in the Hull Lines Design Process Proceedings CADMO. RAVEN, H.C. 1988 Variations on a theme by DAWSON Proceedings 17th Symposium on Naval Hydrodynamics, The Hague, The Netherlands. [Discussion and Closure follow] 80 Shuttle Tanker Design Problems

Discussion Siu C. Fung, Member [The views expressed herein are the opinions of the discusser and not necessarily those of the Department of Defense or the Department of the Navy.] This paper presents an interesting case study for hull form improvements using the state-of-the-art technology CFD- Code Dawson. As one of my recent designs also involves resistance predictions using numerical flow codes, I would like to take this opportunity to raise a couple of questions regarding the presented design procedure: 1. I am a little bit confused with the sectional area curves presented in Fig. 1. Except the bow section, all the three sectional area curves are apparently identical to each other (station 19 and aft). It seems to me that both the "lengthened ship model" and the "final lengthened ship model" were stretched (geosims) from the original design rather than just adding an additional length to the parallel midbody. 2. The "final lengthened ship model" clearly has a longer bulbous bow than the "lengthened ship model," see Fig. 1 of the paper. The illustrations seem contradicting to the text "the bulbous bow length was further reduced." 3. The predicted DAWSON pressure distributions and flow lines for the different configurations were well presented by the authors. However, resistance predictions for the different designs were not presented. Since the first bulb modification is so different from the original design, I have no doubt that the authors only have to base on the pressure distributions and flow lines to conclude that the modified design could be better than the original one. However, I question the wisdom of drawing such kinds of conclusions on the second bulb modification which is a rather local modification when compared to the first modification. I would like to know whether there was any attempt by the authors to use the DAWSON predicted wave-making resistance to back up their claims. 4. I absolutely agree with the authors that after years of development, CFD-Codes are still generally "good in a qualitative sense (relative figures) but not in a quantitative sense (absolute values)." Our recent numerical flow code studies also agree with the authors' claim that "The predicted wave resistance is basically rather sensitive to numerical inaccuracies." In other words, the boundary conditions for the mathematical model and the panels have to be set up correctly and quite often they pretty much rely on the user's experience. As a matter of fact, we have recently used three different DAWSON CFD-Codes to assist in an auxiliary hull form design; we have had great difficulties to establish the free surface panels, particularly in the transom area. The different methods approached by us were rather "numerical" than "physical." Since the presented design has a very low design speed and deep transom immersion, I wish the authors could share their experience on how to set up their free surface panels. 5. I would like to use my recent bulbous bow design studies to illustrate the lesson I learned from the CFD-Code. Figures 11 and 12 depict the pressure distribution and flow line predictions for the two different bulbous bow designs. Despite the difference in their flow lines and pressure distributions, I failed to draw any conclusions like the authors did. I then further pursued the wave-making resistance predictions. The computed results indicated that the second design should have a better performance than the original bulb design. However, our model test results failed to validate this claim. Finally, I would like to make myself clear that I am not denying the achievement of flow codes. I firmly believe that CFD-Code can be a very useful design tool--if we can establish the numerical model of a given design to a stage in which it is advanced enough to duplicate an existing model test result. From that point the confidence level on hull form improvement will be established. Hull form improvements may not always be achieved if we merely base on the flow lines and pressure distributions from the CFD-Codes. John P. Hackett, Member I would like to thank the authors for a fine paper. And I also thank Astilleros Espanoles' willingness to allow the authors to share their experience. I have the following questions. Was DAWSON run for the final bow configuration? If so, what did the DAWSON pressure field output look like? What do the streamlines look like for this configuration? If DAWSON was not run, why? Would the authors kindly include in their written response to this discussion some figures showing the streamline pattern in the bow region around the thrusters and the STL? Can the results from DAWSON be used to determine alignment of bow thruster fairings instead of paint streak model tests? What about stern thruster fairing alignment? The hulls in Fig. 1 of the paper appear to be mislabeled. The second hull appears to be the 126 500 dwt tanker and the third hull the 125 000 dwt tanker. If possible, could the authors in their written response include a comparison of sea trial versus model test performance results for the 126 500 dwt tanker? Ingalls Shipbuilding has very recent experience in the use and advantages DAWSON can offer. Ingalls is in the process of designing and has developed lines for a family of tankers called the Gulf Island Class. The first vessels are 40 000 to 45 000 dwt product tankers. This same hull form will support a bulk carrier and chemical tanker series. The Gulf Island Class of ships has design speeds from 14.5 knots to 16 knots. Initial power estimates were made using the BSRA tanker series. MARIN was then consulted to provide an estimate of powering and for an opinion on possible improvements to the lines. These recommendations were incorporated and a model tested. I was at MARIN to oversee the model testing. Before testing started in earnest, a series of speed runs were conducted to allow me to observe the wave pattern. I felt there was a less than optimum wave pattern in the 14-15-knot speed range. The wave pattern at 16 knots, however, was quite favorable. The initial resistance and propulsion test indicated a slightly higher resistance than estimated by MARIN between 14 to 15 knots. It should be noted, at speeds below 13 knots and at 16 knots and above, the model performance was superior to what Marin had estimated. At that time, we turned to DAWSON to assist in evaluating possible hull form changes to improve the vessel's performance. DAWSON was run for the initial lines and two bulbous bow configurations, one with a stern variation. DAWSON replicated the wave observed during testing at 15 knots and the improved wave pattern at 16 knots. The first change investigated was to the bulbous bow with some softening of the forward shoulder. DAWSON showed a slight reduction in the depth of the wave trough at the bow shoulder. There was no change in the bow wave height. The stern shoulder wave trough deepened, and the amplitude of the stern wave increased. The pressure gradient at the forward shoulder also showed some improvement, however, at the expense of a slightly greater negative pressure at the forward bilge near the bulbous bow. Shuttle Tanker Design Problems 81

Cp ~, 0.70 0.00-0.25 Fig. 11 Pressure distribution and flow line predictions of bulbous bow design #1 Cp + 0.70 0.00-0.25 Fig. 12 Pressure distribution and flow line predictions of bulbous bow design #2 The second hull form modification had the bulbous bow extended farther forward, with softer forward shoulder than the initial hull. In addition, some minor adjustments were made to the stern. The DAWSON results showed the height of the bow wave was reduced. The bow shoulder wave trough showed minor improvement over the initial hull. The stern shoulder wave trough showed the deeper trough identified with the first hull modification. The stern wave results were improved over the first hull modification, though somewhat higher than the initial hull. The pressure distribution results showed an improvement at the forward shoulder and returned to the initial hull's values at the forward bilgebulbous bow region. The stern bilge region showed a minor improvement in the pressure field. It was judged that a small (3 to 5 percent) improvement in powering performance could be expected with the second hull form. Therefore, the model was modified and retested. The results of the tests showed improvements of about 4.5 percent at 14 knots, 3.3 percent at 15 knots, and 82 Shuttle Tanker Design Problems

2.3 percent at 16 knots. All other speed ranges also showed improvements. MARIN has judged our hull form, with the help of DAWSON, as superior. DAWSON indicated an improvement could be expected; however, the determination of the magnitude of said change was still judgmental. Model tests confirmed the improvement and quantified the magnitude of the change. DAWSON has demonstrated its potential for providing beneficial data in a very short period of time; however, it does take experience to correlate the information it produces with how the model will actually perform. DAWSON data can be displayed on a color monitor and the reference point of observation changed in real time. This presentation medium allows the naval architect to zoom in, rotate, and study the DAWSON data. I find this presentation method easier to interpret than the pressure and streamline vector presentation method. Again I would like to thank the authors for a fine paper. Chel Stromgren, Newport News Shipbuilding The authors have done a good job in describing a typical design problem and the important role in which computational fluid dynamics (CFD) can be used to reach a solution. They have presented one major conclusion about the benefits of the use of CFD in design. Specifically, that codes such as DAWSON, can save a great deal of time and expense during the design stage. There is, however, another important advantage in the use of CFD that should be emphasized. By using CFD codes designers not only can reach solutions more quickly, but also have the potential to reach better solutions. There are two components that are involved in this conclusion. The first is that CFD codes can provide design data that are typically unavailable from standard model tests. In this case, the designers were provided with hull pressure data and detailed fluid flow particulars. This data, which would have been difficult to obtain otherwise, were used extensively in reaching a solution. The other facet of CFD codes that allows designers to produce better solutions is the relative speed and ease of use. If the testing had been limited to traditional model tests, only a small number of configurations could have been tested. But, because it was relatively easy to modify and rerun CFD models, the designers were able to test a variety of ideas and converge on the best solution. The result was probably a better design than would have been obtained had only traditional methods been used. As CFD codes become more advanced, the way in which hydrodynamic problems of this sort are analyzed will change dramatically. These tools allow a designer to quickly analyze many design options, some of which would not have ordinarily been considered. The result will be new and better solutions to many design problems. Donald McCallum, NAVSEA [The views expressed herein are the opinions of the discusser and not necessarily those of the Department of Defense or the Department of the Navy.] [Oral.] I'm looking at Fig. 10, I guess. As I understand it, it says the previous ships are the small dots. The best performer actually was the previous ships. I guess the question I have is where do the data come from? It seems as if you had a bad design and you improved it by the bulb, but why didn't you get as good as the previous ships? Is that still a question? It says previous ships, original lines, and modified lines. The lowest curve is the previous ships. I'm wondering why did you deviate from previous practice? The previous ships-- are they some kind of historic series or something which you're using as a database? I guess it's just a question of why you didn't achieve the performance of the previous ships. Authors' Closure We would like to thank all discussers for their interest in our paper and appreciate their useful and positive contributions. The good experiences of some of the discussers with MARIN's potential flow code DAWSON was of significant influence on the motivation of the authors to continue with the further development of these modern design tools. Mr. Stromgren presents an interesting point with regard to " the optimization of hull forms prior to model tests. Since our paper only describes a specific problem, which had to be solved, minor attention was paid to the other outstanding possibilities of the DAWSON code. The most important advantage, as described by Mr. Stromgren, is, of course, the potential of the code to optimize a hull form in a very short period of time. Moreover, it is possible to run a number of alternatives within one to two weeks prior to the selection of the hull form used for the manufacture of the ship model. To further optimize and shorten this process, MARIN is developing an interactive geometry modeling system (GMS) with a fully automated panel generation code, enabling the institute to calculate a series of hull form proposals per day. This new process enabling the designers to easily try new ideas certainly will lead to further improved hull forms. Finally, the authors fully agree with Mr. Stromgren's conclusion that the way in which hydrodynamical problems of this kind are analyzed has dramatically changed with the introduction of these new tools, and is going to change further in the future. In reply to the positive comments of Dr. Hackett, the authors wish to state the following: 1. No DAWSON calculation was made for the final hull form for 2 reasons: (a) The ship model with the hull form of the second alternative was already under construction when it appeared that due to constructional problems with regard to the most forward located thruster tunnel the hull form had to be modified. The selected shape of this forward part of the ship was in between alternative I and II; therefore a pressure distribution in between the pressure determined by means of the DAWSON calculations carried out for said two alternatives was assumed. To keep the original model test schedule, the ship model had to be modified immediately. (b) Costs reduction. 2. A comparison of the paint smear test in way of the bow thruster tunnels and the DAWSON calculations of the same region is presented in Fig. 13 herein. For full block ships, showing relatively short waves, the alignment of the bow thruster tunnel using the results of the potential flow calculations is possible and as shown above quite reliable. For high speed ships with relatively long waves related to the Froude number of the ship, the alignment of the thruster tunnel is more sensitive for the shape of the hull form and hence has to be treated more carefully. Some further correlation between model test results and the new nonlinear potential flow code RAPID is needed. The alignment of the stern thruster for a single-screw vessel by means of a potential flow calculation is felt to be insufficiently accurate. Given the unknown structure and expected increased thickness of the boundary layer in the afterbody region and the influence of the working propeller, the actual flow can be quite different when compared to the potential flow direction calculated. Most probably the results of viscous flow calculations, expected to become more common practice Shuttle Tanker Design Problems 83

- - STREAMLINES DERIVED FROM PAINT TEST BODY PLAN, STEM AND STERN PROFILES AND SECTIONAL AREA CURVE OF SHIP MODEL - - - - FLOW DIRECTION DERIVED FROM DAWSON J. J ORIGINAL SHIP MODEL J J J 0 2 20 19 j, Star 0 - A P $1a1.20 - F,P LENGTHENED SHIP MODEL 18.5 19 19.5 Fig. 13 2O in the coming years, can be used for the alignment of stern thruster tunnels, ensuring accurate and reliable results. 3. In Fig. 14 the corrected Fig. 1 of the paper is presented. In the original figure, indeed, a mistake was made. 4. With regard to the presentation of a comparison of sea trials versus model tests, the authors are unable to present data for confidentiality reasons. On the other hand, it might be interesting to know that Astilleros did not have to pay any penalty due to speed problems for the last ten years. This implies that the sea trial results are well in line with the extrapolated model test results. Finally, the authors are very pleased with the description by Dr. Hackett of the process followed at MARIN to optimize the hull form of the Gulf Island Class tanker. The authors were pleased to hear from Mr. Hamalainen (oral) of Kvaerner Masa, Finland, that the paper was interesting to him. Moreover, his favorable experiences with the use of potential flow codes in the daily design routine in his yard was interesting information. Regarding the remarks, comments, and experiences of Mr. Fung, the authors would like to respond as follows: 1. Please find presented here in Fig. 15 a comparison of the curves of sectional areas related to the three configurations tested. This comparison clearly shows that the parallel midbodies were lengthened. 2. Your comments regarding the contradiction between the text and the body plans presented in Fig. 1 is correct (see Fig. 14). 3. Of course, the authors made an attempt to predict the wavemaking resistance. However, as described earlier in Mr. Ravens paper [1] (additional reference follows this closure) presented at the ONR Symposium, 1990, a negative wavemaking resistance value is often calculated by the DAWSON program for these full block ships, which is inherent to the linearization of the code. With regard to the results as presented in the paper, the authors are convinced that the problems encountered are related to the viscous resistance component (flow separation and vortex shedding) rather than to the wavemaking resistance component. 4. As a matter of fact, the paneling of hull and free surface may affect the resistance estimates, but they have much less effect on the pressure distribution, wave pattern, and streamline direction. It is this information that is used in our design Slat 0 = A.P Slat. 0 = A.P FINAL LENGTHENED SHIP MODEL o 2 2O 19 Fig. 14 Star, 20 - F,P, Star, 20 - F.P. procedure, which, therefore, is relatively insensitive to the particular paneling used. For the present cases, the deep transom immersion and low speed make it useless to model the flow as if it clears the transom; a dead-water area will instead be present, which cannot be properly modeled in a potential flow. In such cases we sometimes resort to adding a smooth extension of the stern that ends in a single point or a line. Physical realism cannot be expected for the local flow prediction. For completeness sake, the paneling of the free surface is presented in Fig. 16. 5. From the plots presented, the precise pressure levels and flow directions are hard to distinguish, and we thus cannot make any judgment of which variation we would expect to be better. We have some experiences in which the predicted resistances incorrectly pointed out one design to be better while actually it was not; but the wave pattern, pressure distribution, and streamline direction then did give the correct indication. One should only confide in the predicted resistance difference if it well agrees with your assessment of the predicted wave patterns and flow behavior. Mr. Latorre's (oral) final conclusions regarding the usefulness of the introduction of these CFD codes on universities as an educational tool for the students are fully supported by the authors. Finally, the authors once more would like to thank the 84 Shuttle Tanker Design Problems

SECTIONAL AREA CURVE OF SHIP MODEL FINAL LENGTHENED SHIP MODEL... LENGTHENED SHIP MODEL.......... ORIGINAL SHIP MODEL f Star 0 = A.P, Star. 20 = F.P. Fig. 15 PANELS ON FREE SURFACE Fig. 16 discussers for their comments and SNAME for the opportunity to present this paper. Additional reference Raven, H.C. 1990 Adequacy of free-surface conditions for the wave-resistance problem. 18th Symp. Naval Hydrodynamics, Ann Arbor, MI. Shuttle Tanker Design Problems 85