SUSAN L. BALES PRACTICAL SEAKEEPING USING DIRECTIONAL WAVE SPECTRA During the past decade, ship dri ers, owners, and de igner ha e realized that knowledge of the prevailing wave environment can be put to good use in improving hip-seakeeping performance. The wave height is important, but ignorance of the wave length and wave direction( ) can ha e equally adverse effects on ship performance. This article identifies some sensiti ities of hips to the wa e pectrum and pro ides resolution requirements for engineering applications . INTRODUCTION For decades, ocean-going Navy ship hulls had been designed merely to optimize calm-water performance. That is, ship hull forms had been developed to en ure maximum speed in calm water. Nearly 20 year ago, an attempt was made to consider hip performance in waves, i.e., seakeeping. Such early attempts resulted in less than optimum ships. Using the best ad ice a ailable at the time, the Navy designed several clas es of hip to maintain speed in a seaway defined by a P iersonMoskowitz Sea State 5 pectrum. Thus, the ships were constrai ned to perform well in about a 3-meter unidirectional seaway with a fixed modal period. The result was that the ships were well tuned for a wave pectrum that may never occur in nature . Actually, lower sea states could (and did) cause much more excessive motions of the ships from other combinations of wave height, period , and direction. About 10 year ago, the Navy decided to try again. By then, naval architect worldwide had adopted the Bretschneider two-parameter spectrum combined with a cosine-squared angular dependence to reflect directional spreading of the seas about a gi en, fixed primary direction. These spectra were initialized with wa e heights and periods deri ed from visual ob ervation from ship of opportunity. Thus, probabilitie of occurrence were developed for the spectra by the joint occurrence statistics of the wave height and period . While this methodology was a va t impro ement, it had the follo ing fau lts: 2. i ual e timate of v a e conditions were/ are bia ed b the capabilit of the ob erver, hip ize hipping lane , etc. 3. Sv ell-corrupted \ ind-generated ea \ ere ignored . In 19T , the a eized an opportunit to addre the e deficiencie. The p tral 0 ean Wa Model (SOWM), ba ed on the theorie of Pier on and hi 01league , I had been made op rational at the Fleet umerical Weather Center b Lazanoff and Ste en on.The SOWM pro ided t\ ice-dail foreca t of directional \ a e pectra throughout the orthem Herni phere. Concurrentl , Fleet commander v ere calling for improved eakeeping performance. Too often, U.S. ships were just not able to keep up \ ith A TO allie and So iet counterpart . Figure 1 illu trate a hip pitching-and-slamming problem that clear! limit fon ard peed . The turning point came from the man re earch program initiated a a re ult of a ork hop held at the a al Academy in 1975. 3 Some of the program are identified in Ref. 3. Of prime intere there i the program propo ed b the En ironment Group, chaired b Cumrnin of the Da id W. Ta lor a al Ship R&D Center. The group \! hi h con i ted of na al architect oceanographer meteorol- 1. Reliable global height and period stati tics were unavailable. Susan L. Bales i head of the Ocean Enironment Group of the Ship Surface D namic Branch , the Da id Taylor aval Ship R&D Center , Bethe da, MD 20084. Figure 1-Pitching motion. keel slamm ing, and , sea spray while refueling in moderate to heaving seas. April 1962. 42 Johns H opkin APL Technical Dige l. \' olume . umber I (/9 ) ogists, and modelers, postulated the concepts that have become, in many cases, the state of the art. For instance, a massive wave-hindcasting effort resulted that has been reported in numerous papers, articles, and atlases (e.g., Ref. 4). At present, the Navy has addressed, to some extent, all of the deficiencies. In ship design, the Bretschneider cosine-squared spectra are still routinely used, but they are initialized by hindcasted significant heights and modal periods. For specific investigations where multidirectional seas are critical, either Cummins' stratified sample of hindcast directional or other SOWM spectra are used. The SOWM has been replaced by a global version (GSOWM) implemented by Clancy, Kaitala, and Zambresky 5 in 1985. However, the Navy is still deriving great benefit from the SOWM hindcasts, and the GSOWM forecasts are now being used operationally with other sensors and models to analyze ship seakeeping behavior during ship deployments. B ~ (l) a. o (l) "D If (l) (/) c o a. (/) ~ 3 (j) ~u (l) a. (/) (l) DIRECTIONAL SPECTRA REQUIREMENTS 6 But more still needs to be accomplished. N. Bales and Walden and Grundman 7 have clearly demonstrat'ed that ship hull forms can be optimized to the seaway. In fact, acceptable operability in northern latitudes can probably be extended upward by several sea states by desensitizing the natural resonances of the ship hull to the estimated prevailing wave conditions. The engineering community has developed a highly sophisticated shipmotion program that requires a directional wave spectrum as the forcing function. Given the wave spectrum, most motions in moderate to heavy sea states are predicted to within ± 10 percent accuracy when compared to towing-tank simulations. The response amplitude operators (defined as the square of the ship-transfer function) are different for every ship and depend on its size, shape, appendages, hull form, and ballast condition. They also vary for every combination of speed and course relative to the prevailing seaway. In Fig. 2, it is obvious that the frigate (about 122 meters long) has substantially more response in pitch (the vertical angular motion about the center of gravity, as seen in Fig. 1) and at higher frequencies than does the aircraft carrier (about 274 meters long). Clearly, the wave spectrum (plotted on the lower portion of Fig. 2) that most closely aligns with the response function peak causes the greatest pitch motion. The wave spectra in the figure are Bretschneider unidirectional spectra for a fixed height but for varying modal periods. A deficiency in ship motion prediction remains in the modeling of wave directionality. The cosine-squared law is still applied to spectra such as those in Fig. 2, occasionally using SOWM/ GSOWM spectra, as mentioned previously. Figure 3 illustrates what is presently used versus what is really needed in order to achieve adequate information about swell-corrupted wind-generated seas. The importance of directionality is clearly illustrated in Fig. 4, where there is about a factor of two between predicted ship-rolling (side-to-side) angles for unidirectional (long-crested) and cosine-squared (short-crested) Johns H opkins APL Technical Digest, Vol ume 8, N umber 1 (1987) > co ~ 1 "D (l) .~ co (l) ~ Wave frequency, W Figure 2-Prediction of ship response, e.g., rms pitch RAO{w} S {w} dw] V2 . [J beam seas (seas traveling 90 degrees to the major ship axis). Another side of the problem lies in the operational application of directional wave spectra. Future tacticaldecision aids require a good in-situ directional wave spectrum. The Navy's Tactical Environmental Support System (TESS) will require a reliable spectrum in order to predict ship responses. Required spectral resolutions are given in Table 1. These resolutions are based on the known variability of the ship's transfer functions, as illustrated for pitch in Fig. 2. Figure 5 shows some generic sensitivities to modal wave period as ship length and hull form are changed. Several approaches could help achieve the data requirements of Table 1. It is not clear that ships at sea can depend on a single land-based forecasting system nor does a single spaceborne system seem to be emerging that could accomplish the requirements. The optimum is to have available data from a variety of sensors and models so that the tactician can select data to appropriate levels of complexity and resolution. Such a strategy might include the following elements: 1. Deploy a disposable wave buoy that, by telemetry, provides acceleration data to be rapidly processed (on board) into wave-height spectra. The current cost is about $4,000 per buoy, including antenna and receiver. Processing requirements are minimal. Total cost could probably be decreased to about $500. 43 Bales - Pracfical Seakeeping Using Direcfional Wa ve Specfra 8~---------.----------.----------, Primary heading Sign ificant wave height Ship speed = 20 knots 6 3.6 meters Long -crested waves en Q) ~ Ol Q) ~ ~ Ol 4 Short-crested c co ern E a: 2 ~------ Symmetric cosine 2 (a ppl ied to two-parameter spectra of Fig. 2) Following seas o 120 60 Ship heading angle to predominant wave direction (degrees) Head seas Figure 4-Comparison of pred icted ro lli ng (side-to-side) motion using Bretschne ider long-crested and short·crested sea models. 12 ~----'-----'-----.-----.-r---'---71 - - Hogben and Lumb (visual ) - - Hindcast climatology Q) > 6 co rn ~ Q. E +-' C co u rn 3 Q) Eco ~ c 180 .~ Swell-corrupted , wind-driven sea (SOWM ) Figure 3-G eneric comparison of cosine·squared directional model against a more representative hindcast directional spectrum , showing a swell-corrupted wind-generated sea. Resonant pitch • and roll (j) 0 5 m 7 9 11 Most probable modal wave period (seconds ) 17 Figure 5-Significant wave height versus the most probable modal wave period showi ng sh ip pitch and ro ll resonant period ranges . Table 1-Required directional wave-parameter resolutions for ship·response prediction. Parameter Resolulion Significant wave height ±O.3 meter for Sea States 4- 7 (1.25-9 meter) Modal ± I second for wa es from 3-24 econds 'V a e period Directional preading ± 7.5 degrees for primary and econdary system (or IS-degree increment about the compass) 44 2. U e the di po able wa e buo together \ ith the shipboard na igation radar (e.g . the SP -64) to get direction information a \' ell. Trizna and hi colleague at the a al Re earch Laborator are currently exploring thi technique. 3. Deploy an Endeco / Data ell or other portable directional wa e buo . Co t range from 1~ 000 to $80,000 and are related to buo ize and th required pectral re olution. 4. Use pace or aircraft remotel en ed data. GeoJohns H opkin APL Techni'al Dige I. \ ·olume . ,\ ·umber I (/9 - ) Practical Seakeeping Using Directional Wa ve Spectra Bales - sat altimeter wave-height data appear to be the only hopeful sign for the United States in the near term, with the delays in the Shuttle Imaging Radar Program and the recent (December 1986) apparent demise of the Navy Remote Ocean Sensing System. It is clear that the United States now lags both Japan and Europe in a national space-oceanography program, particularly one to support military environmental-monitoring requirements. Routine deployment of aircraft sensors has not yet been established as a requirement. But from the Navy battle group viewpoint, aircraft sensors may offer a viable alternate to some spaceborne systems, particularly for timely local weather forecasting. 5. Use properly validated GSOWM and other landbased forecasts of directional wave spectra. In a statistical sense, we have found the SOWM I GSOWM data to be quite reliable, and, on some occasions, individual forecasts could be used to provide meaningful guidance to the ship driver. Figure 6 is a speed polar-graph for a ship that encountered high waves in the Northeast Pacific. Here, the concentric circles represent ship speed in increments of 5 knots. The radial lines represent ship-to-wave relative headings. The @ represents the operating condition of 17 knots in port bow seas, a condition in which severe damage was in- The usefulness of Fig. 6 is obvious, but the forcing function or directional seaway must be well defined to predict ship motion. Figure 7 illustrates a North Pacific case where buoy and GSOWM data are compared and 3.5r----.---.r---.----.r---.----.----~~ (a) ~CIl .- ~ 3.0 1986) - - Buoy: significam wave height = 4.1 meters -g 2.5 ~ 8 (52.5°N,137.5°W 1200 GMT, April 9, 1986) - - GSOWM : significant wave height = 4.08 meters ~ ~ 2.0 +-, ' UN Q) CIl Q.(j)1.5 CIl +-' Q) Q) i6 E 1.0 S 0.5 o Probable (95 percent) damage D Possible (5 percent) damage o No damage but occasional wet (stern) IVDS doors ® curred by the 8.2-meter waves. The wave forecasts agreed well with the ship observations. The red area represents conditions posing a 95 percent probability of damage, while the yellow area represents conditions posing more than a 5 percent probability of damage. If this wave forecast and speed polargraph had been available to the captain, he would have quickly seen that a slight (30-degree) change of course to port would have drastically reduced the potential of damage to the ship. It is planned to incorporate these types of tactical decision aids into the Tactical Environmental Support System. Operating condition incurred damage to the top mast structure ,315 Waves (SOWM: 8.2 meters, 14 seconds 270 r--r--r-~--~~ oL-~---i---L~~===c===- __~ 1.6r----1r---~1r---~1r---~1----~1r---~1----~1~ ~ .~ (b) - - Buoy: significant single amplitude_ pitch = 1.0 degree 1.4- Q)~ ~ ~ 1.2co c ;; a - - - GSOWM: significant single amplitude pitch = 0.8 degrees (significant single amplitude = 2 x RMS amplitude ~ ~ 1.0Q.CIl ~'iO.8- Q) - Q) Q) +-' § 0>0.6 fa - Q) U"O ~ ~0.4f- - 100 (e) ~ 'en c Q) - "0 CIl _ "0 - - Buoy: significant single amplitude = 4.9 degrees - - GSOWM : significant single amplitude = 4 .0 degrees 80 co C ~ ~ 60 c:::J 40 Q.CIl CIlN ~ CIl Q) Q) a ~ OJ Q) u"O c~ Q) 180 Ship speed / heading plane Figure 6-Speed polar-graph of a ship operating in the northeastern Pacific on March 8, 1974, where a heading change of about 30 degrees would have reduced the probability of damage by 90 percent ; speed changes would have far less impact, if any. Johns H opkiIlS A PL Technical Digest , Vo lum e 8, N um ber I (1 987) 20 "0 ex:: 0.4 0.8 1.2 1.6 2.0 2.4 Wave encounter frequency (RPS) 2.8 Figure 7 -Comparison of GSOWM and buoy measurements and motion predictions for an aircraft carrier. The agreement between spectra permits a good forecast of ship motion. 45 Practical Seakeeping Using Directional Wa ve Spectra Bales - the resulting pitch and roll motion is computed for an aircraft carrier. The Delft Disposable Wave Buoy, a standard for developing wave-height spectra, was deployed along a ship's route. The GSOWM forecasts were taken for the clo est temporal and patial GSOWM gridpoint. The point spectra (Fig. 7a) agree well with the forecasts, and the resulting predicted pitch and roll responses (Figs. 7b and 7c) a re al 0 comparable to the predictions. Figure 8 provides a com pari on during the arne ship transit when the measured and foreca t spectra differ. The GSOW M spectrum contain substantially more energy than that derived from the buoy vertical-acceleration measurement, resulting in a ubstantial difference between the two predicted ship re ponses. The e differences could be even larger for a frigate, which ha more 4 .0 (a) "?: _ -2 - 'iii ~ c 3 .0 1:) 53 .9 °N , 153 .7 ° W 0200 GMT, Ap ril 12, 1986 Buoy: significant wave height = 3.4 meters 3.5 0 U ~ ~ 2 .5 +-' ' UN ~ ~ en ~ 2.0 - E1.5 ~ 55.0 0 N, 152 .5°W 0000 GMT, April 12, 1986 GSOWM : significant wave height = 4.7 meters Cll- ~ re pon e in the modal a e-frequency region of the wave pectrum . Analyticall deri ed hip-re ponse transfer function were u ed to de elop both Fig. 7 and 8. Wave forecasts can al 0 be u eful e en when they are at ariance with mea urement of the local area. For example, GSOWM wa e height can be scaled by Geosat altimeter a e height. During recent ship trials in the Northwe t Pacific, the foreca t wa e period and directions agreed well with tho e ob erved by experienced peronnel and with tho e deri ed from ship-motion measurement . Ho e er, the forecast wave energy of the GSOWM directional pectra appeared to be too high. Figure 9 pro ide a compari on of these data during December 16-18, 1986. The darkened circle represent the Geosat er us GSOWM ignificant wa e heights. The colored line approximate the error associated with the altimeter height a deri ed by Monaldo. 8 In general, the GSOWM data for the period indicate a 30 percent higher ignificant wa e height than what was measured. If the two point with a di tance greater than 100 nautical mile bet een the grid point and the 0 erflight point are excluded, the GSOWM data indicate a 22 percent greater height. The conclu ion here i that in orne case the GSOWM pectra can be caled b imple height data to de elop a directional wa e pectra. Other comparion of forecast and Geo at data ha e re entl been reported b Pickett. 9 1.0 0.5 Distance between Geosat measurement and GSOWM grid po int o~LJ--~~~~~==~--~--~ 2 . 0r---'-~-.----.----.----.---.----._. (b) "?: .iii c ~ -1.6 en - 1:) ~ C 0 +-' - Buoy: significant single amplitude pitch = 0 .8 ° - GSOWM: significant single amplitude pitch = 1.3 ° (significant single amplitude = 2 x RMS amplitude ~ ~ 1.2 en en ' ... Nen Q. 0) 0) ::::l 0> c a < 50 nm i b50- 100nm i c 100-200 nm i I Geosat spatial and temporal error range = _ 0.35 m -.:; 12 0) 0) O) 4 -5 hr 5-6 hr 6-7 hr 3 14 r----.-----.----~----._--~,_--_.--__, ~ U1:) c- 1 2 Examp le: (a, 2 ) ind icates a distance of <50 nmi w ith a 5-6 hour t ime difference ~ 0.8 o T ime difference Geosat - GSOWM 0.4 = b, 3 1216 0619 £ B b. 3 ) (a , 1 1216 1216 0623 1601 0:: 0) 200r---.-.--.----.----.----r---~--~_. (e) "?: .iii ~ 1:)- 160 ~-g Buoy : significant single amplitude = 4 .3 ° 0 Cll ~ GSOWM : significant single amplitude = 6 .6° 6U f 0> +-' Cll 4 en 0 0) ~ 80 <.9 OJ 2 0) 40 0 o a:: o~~~~~ o ________ ________ ____ ~ ~ 0.4 0.8 1.2 1. 6 2.0 2.4 Wave encounter frequency (RPS ) 0 ~ 2 4 6 10 12 8 GSOWM sign if icant wave he ight (feet ) 14 2 .8 Figure a-Compari so n of GSOWM , buoy measurements and motion predictions. The disagreement between spectra provides a poor forecast of ship motion . 46 6 .iii , en c~ 0) 1) C Q.en 0) c 1218 1634 C ... N C 8 ~ +-' ~ ~120 en > Cll U - en +-' - f Figure 9-Comparison of Geosat and GSOWM signi ficant wave heights in the northw estern Pac ific for December 16-18, 1986. The comparison gene rally show s that the GSOW M fo recas ts are somewhat high . 1217 1707 = date and time of Geosat; Decem ber 16-18, 1986. John H opkins APL Technical Dige I, Volume , umber I (/9 ) Bales - CONCLUSIONS This article has established some of the practical applications for which accurate directional wave spectra are required. The spectral resolutions required for practical ship applications have been identified, though it is expected that they will be achieved only in incremental steps. From the engineering viewpoint, climatological data are generally sufficient. They can be developed by hindcasting techniques or by global wave measurements, although the latter can probably be achieved only by means of spaceborne sensors. Spaceborne altimeters such as those on Seas at and Geosat appear to provide adequate estimates of wave heights. These height estimates can be used to "tune" land-based global models or seabased regional models and measurements. It is not yet clear that space sensors such as the synthetic aperture radar will provide the full directional wave-energy spectra. The article by Beal in this issue describes some of the limitations of synthetic aperture radar at the shorter wavelengths of interest. Further measurements and intercomparisons in realistic sea states should help resolve this question. Naval engineers must take advantage of every technological edge in order to gain even the slightest improvement in seakeeping. A better knowledge of the prevailing wave environment will improve operability. Additionally, real-time measurement of global wave conditions could vastly improve forecasting products available to the Fleet. But to counterbalance the vulnerability of satellites, it is essential that both in-situ and shipborne sensor development be accelerated. With the recent cancellation of the Navy Remote Ocean Sensing System program, these alternative approaches take on added importance. With regard to national responsibilities, joint agency programs are required in the areas of sensor development, data assimilation into models, and model valida- Johns Hopkins APL Technical Digesf , Volume 8, N umber I (/987) Practical Seakeeping Using DirecTional Wave Spectra tion. A critical problem here is in identifying areas of responsibility. Such national programs should be reflective of NASA's fundamental science interests, the Navy's interest in improving Fleet readiness and operability, and NOAA's interest in supporting commercial fishing, offshore industry, and the civilian population in general. Such programs can be established only through careful interagency negotiation and cooperation. REFERENCES 1 W. J. Pierson, L. J. Tick, and L. Baer, "Computer Based Procedures fOJ Preparing Global Wave Forecasts and Wind Field Analyses Capable of Using Wave Data Obtained from Spacecraft ," in Proc., 6fh Naval Hydrodynamics Symposium, pp. 499-532, Washington , D.C. (1966) . 2 S. M. Lazanoff and N. M. Stevenson, An Evaluation of a Hemispheric Operational Wave Specfral Model , Fleet umerical Weather Center Technical Note 75-3 (1975). 3 Seakeeping in the Ship Design Process, Research and Technology DirecfOrafe Reporf of fhe Seakeeping Workshop, Naval Sea Systems Command , Washington, D.C. (1975). 4 S. L. Bales, " Development and Application of a Deep Water Hindcast Wind and Wave Climatology," in Proc., imernafional Symposium on Wave and Wind Clima fe Worldwide, Royal Institution of aval Architects, London (1984). 5 R. M. Clancy, J. E. Kaitala, and L. V. Zambresky, " The Fleet Numerical Oceanography Center Global Spectral Ocean Wave Model," Bull. Am. Mefeorol. Soc. 67, 498-512 (1986). 6 N. K. Bales, "Optimizing the Seakeeping Performance of Destroyer-Type Hulls," in Proc., 131h Symposium on Naval Hydrody namics, pp. 479-504 (1980). . 7 D . A. Walden and P. Grundman, "Methods for Designing Hull Forms with Reduced Motions and Dry Decks," Nav. Eng. 1. 97, 214-223 (1985). 8 F. Monaldo, Expecled Differences in lhe Comparison of Wind Speed and Sig- nijicam Wave Heighl Made by Spaceborne Radar Allimelers and Buoys, JH UlA PL Technical Report SIR86U-017 (1986). 9 R. L. Pickett, "Ocean Wind and Wave Model Comparison for Geosat," in Proc. , imernalional Workshop on Wave Hindcasfing and Forecasling (1987). ACKNOWLEDGMENTS- I wish to acknowledge the support received from the Ocean Environment Group, Surface Ship Dynamics Branch , of the David W. Taylor aval Ship R&D Center (DT SRDC) under a variet y of efforts. The results reported are primarily derived from the Surface Wave Spectra for Ship Design Program, P .E. 62759 , administered by the aval Ocean R&D Activity for the Office of Navy Technology, for which DTNSRDC has been the lead laboratory . I am grateful for the special efforts of A. L. Silver , W . L. Thomas III, and R. J. Bachman of DTNSRDC for their preparation of Figs. 7, 8, and 9 . 47