A NEW METHOD O THRUSTER CONTROL IN POSITIONING O SHIPS BASED ON POWER CONTROL Asgeir J. Sørese*, Alf Kåre Ådaes*, Thr I. sse** ad Ja-Peter Strad** *ABB Idustri AS, Hasleveie 50, P. O. Bx 6540 Rdeløkka, N-050 Osl, NORWAY (E-mail: asgeir.srese@ia.abb. ad alf-kare.adaes@ia.abb.) ** Nrwegia Uiversity f Sciece ad Techlgy, Departmet f Egieerig Cyberetics, N-7034 Trdheim, NORWAY (E-mail: tif@itk.tu. ad ja@itk.tu.) Prceedigs f the 4th IAC Cferece Maeuvrig ad Ctrl f Marie Craft (MCMC'97) Abstract. Psitiig systems iclude differet ctrl fuctis fr autmatic psitiig ad guidace f marie vessels by meas f prper acti f the thruster ad prpeller devices. These may be ctrllable pitch prpeller (CPP) with fixed speed, ctrllable speed with fixed pitch prpeller (PP), r ctrllable pitch ad speed i cmbiati. The psitiig ctrller cmputes cmmaded frces i surge ad sway ad mmet i yaw. The thruster allcati algrithm determies the crrespdig frce ad directi each thruster ad prpeller device must prduce i rder t fulfil the psitiig ctrller cmmads. Cvetially, the fial pitch r speed set-pit sigals are determied frm statiary prpeller frce t speed/pitch relatis based data frm the prpeller maufacturer. Hwever, these relatis are strgly iflueced by the lcal water flw arud the prpeller blades, hull desig, peratial philsphy, vessel mti, waves ad water currets. I this paper, a ew methd based trque ctrl ad pwer ctrl f the prpeller ad thruster devices is itrduced. Istead f calculatig the prpeller speed ad pitch set-pit sigals based statiary fuctis, the prpeller frce t mmet ad pwer relatis are used. This methd gives a sigificat imprvemet i the perfrmace ad the stability f the electrical pwer plat etwrk, while the psitiig accuracy ad badwidth are imprved. Simulatis demstrate the perfrmace f the prpsed ctrller schemes. Key Wrds. Psitiig systems, pwer ctrl, trque ctrl, thruster ctrl, marie systems.. INTRODUCTION Psitiig systems such as dyamic psitiig (Dyps) systems, thruster assisted psiti mrig (Psmr) systems, ad autsail systems fr marie, il, ad gas vessels have bee subject t a icreased fcus with respect t fuctiality, perfrmace ad safety due t a icreased umber f applicatis fr ffshre il explrati ad explitati at margial fields ad at deep water depths i hstile evirmets. The strg requiremets t vessel perfrmace ad verall safety have als resulted i icreased fcus the ttal vessel ccept ad the iteractis betwee the differet equipmet ad systems istalled. lexibility i perati has eabled electrical pwer geerati ad distributi systems fr prpulsi, psitiig, il prducti, drillig, ad ladig, where all equipmet ad ctrl systems are itegrated it a cmm pwer plat etwrk ad autmati etwrk. Psitiig systems have bee cmmercially available fr marie vessels sice the 960s. Hwever, it is first i the 990s that fully itegrated electrical pwer, autmati ad psitiig systems have becme cmmercial available. Up t w, almst all atteti i the develpmet f psitiig systems has bee directed t imprvemet f the psitiig ctrl strategies, fr refereces see sse (994), ad Sørese et al. (996). Less atteti has bee made t the thruster allcati ad the iteral thruster ctrl with sme few exceptis, see Blake (98), Grimble ad Jhs (988), sse (994), Healey et al. (995), Sørdale (996 ad 997) ad Berge ad sse (997). The psitiig systems iclude differet ctrl fuctis fr autmatic psitiig ad guidace. The psitiig ctrller cmputes cmmaded frces i surge ad sway, ad mmet i yaw. The thruster allcati algrithm determies the crrespdig frce ad directi each thruster ad
prpeller device must prduce i rder t fulfil the psitiig ctrller cmmads. The prpeller ad thruster devices ca be ctrllable pitch prpeller (CPP) with fixed speed, ctrllable speed with fixed pitch prpeller (PP), r ctrllable pitch ad speed i cmbiati. The prpeller pitch rati is measured as the agle at a radius 0.7 R f the prpeller blade r as the rati f travelled distace per revluti divided by the prpeller diameter. Cvetially, the fial pitch r speed set-pit sigals are determied frm statiary prpeller frce t speed/pitch relatis based ifrmati abut thruster characteristics ad bllard pull tests prvided by the thruster maufacturer. These relatis may later be mdified durig sea trials. Hwever, they are strgly iflueced by the lcal water flw arud the prpeller blades, hull desig, peratial philsphy, vessel mti, waves ad water curret. I cvetial psitiig systems, variatis i these relatis are t accuted fr i the ctrl system resultig i reduced psitiig perfrmace with respect t accuracy ad respse time. I additi, the variatis may als lead t deterirati f perfrmace ad stability i the electrical pwer plat etwrk due t uitetially peaks r pwer drps caused by lad fluctuatis the prpeller shafts. The upredictable lad variatis frce the peratr t have mre available pwer tha ecessary. This implies that the diesel geeratrs will get mre ruig hurs at lwer lads i average, which i terms gives mre tear, wear ad maiteace. I this paper a ew methd based trque ctrl ad pwer ctrl f the prpeller ad thruster devices is itrduced by usig prpeller frces t trque ad pwer mappigs. 2. MATHEMATICAL MODELLING 2. Prpeller Characteristics The prpeller thrust ad trque T are frmulated accrdig t : = ρd 4 K T, () T = ρd 5 K Q, (2) where ρ is the water desity, D is the prpeller diameter, ad is the prpeller speed (revlutis per secd). The expressis fr K T ad K Q are fud by s-called pe water tests, usually perfrmed i a cavitati tuel r a twig tak. These -dimesial thrust ad trque I this paper detes thrust ad T detes trque i rder t crrespd with the meclature i the el. pwer literature. cefficiets are described by the fllwig parameters (Osterveld ad va Ossae, 975): K f J P A E D A Z t T =,,,, R,, (3) c K f J P A E D A Z t Q = 2,,,, R,, (4) c where P/D is the pitch rati, A E / A is the expaded-area rati, Z is the umber f blades, R is the Reylds umber, t is the maximum thickess f the blade secti, ad c is the chrd legth f the blade secti. J is the advace rati defied as: Va J =, (5) D where V a is the iflw velcity t the prpeller. The pe water prpeller efficiecy i udisturbed water is give as the rati f the wrk de by the prpeller i prducig a thrust frce divided by the wrk required t vercme the shaft trque accrdig t: Va J K T η = =. (6) 2πT 2π K K T, K Q, ad η 0 curves fr differet pitch ratis fr a Wageige B-screw series based Table 5 i Osterveld ad va Ossae (975), with R = 2x0 6, Z = 4, D = 3. m, ad A shw i ig.. 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0. Q E / A = 0.52 are 0 0 0.2 0.4 0.6 0.8.2 ig. Ope water K T (slid), 0 K Q (dash-dt) ad η (slid) as a fucti f advace rati J fr P/D=0.7, 0.89 ad.. 2.2 Prpulsi Efficiecy The differece betwee the ship speed U ad V a ver the prpeller disc is called the wake. r a ship mvig frward, V a is less tha U sice the afterbdy flw chages its magitude betwee ship
speed ear the vessel ad zer far frm the vessel. This statiary relatiship fr axial water iflw ca be mdelled as (Walderhaug, 992): V = U ( w + w + w ) = U( w), (7) a w P v where w is the wake fracti umber, typically i the rage f 0 < w < 0.4. w w is the wake fracti caused by the wave mti f the water particles, ad ca i irregular waves with N cmpets at a average immersi equal t z be estimated t: u wx w w = U N ei ai U e kiz = ω ζ (8) si( ω ei t + k i x ) where ω ei i= 2 ω0 = ω iu 0i + g is the circular wave frequecy f ecuter fr wave cmpet i, ζ ai is the wave amplitude, k i is the wave umber, ad g is the gravity cstat. w p is the wake fracti caused by s-called ptetial effects fr a hull advacig frward i a ideal fluid, ad w v is the wake fracti caused by viscus effects due t the effect f budary layers. Oe shuld tice that (8) must ly serve as a rugh estimate, sice it des t accut fr the iteractis betwee the vessel mti ad the actual wave particle flw. By applyig a mmet equati t a ctrl vlume surrudig the ilet flw, the thrust ca be related t the rate f chage f mmet t the ctrl vlume. This will give a dyamic equati fr the advace speed, see Healey et al. (995) ad the refereces therei. The sucti f the prpeller geerally reduces the pressure at the ster resultig i icreased drag D accrdig t: D = ( t d ), (9) where t d is the thrust-deducti cefficiet, typically i the rage f 0 < t d < 0.2 caused by pressure reducti due t ptetial effects, viscus effects, waves, ad appedices (Walderhaug, 992). I sme extreme cases t d may becme egative. The thrust cefficiet behid the hull is rmally assumed t be uchaged cmpared t pe water, while the trque cefficiet will be affected by the chage i iflw at the ster. This is accuted fr by the relative rtative efficiecy: J K T η 2π K B QB K Q ηr = = =. (0) η η K The verall prpulsi efficiecy ca w be fud as the rati betwee the useful wrk de by the QB prduct f drag ad ship speed divided by the wrk required t vercme the shaft trque: ηr = DU = ηhηηrηm, () 2πT where η = t d h w is defied as the hull efficiecy ad is accrdig t Newma (977) i the rage f.0-.2. η m is the mechaical efficiecy typically i the rage f 0.8-0.9. 2.3 Prpeller Lsses I additi t the mdelled prpeller lsses caused by axial water iflw, several ther effects will ctribute t reducti f prpeller thrust ad trque (Leh, 992, Kaase, 996) such as: Water iflw perpedicular t the prpeller axis caused by curret, vessel speed r jets frm ther thrusters will itrduce a frce i the directi f the iflw due t deflecti f the prpeller race. This is fte referred t as crsscuplig drag. r heavy laded prpellers vetilati (air sucti) caused by decreasig pressure the prpeller blades may ccur, especially whe the submergece f the prpeller becmes small due t vessel s frequecy wave mti. r extreme cditis with large vessel mtis the i-ad-ut-f-water effects will result i sudde drp f thrust ad trque fllwig a hysteresis patter. Bth thrust reducti ad chage f thrust directi may ccur due t thruster-hull iteracti caused by frictial lsses whe the thruster race sweeps alg the hull. This is referred t as the Cada effect. Thruster-thruster iteracti caused by ifluece frm the prpeller race frm e thruster eighburig thrusters may lead t sigificat thrust reducti, if t apprpriate precautis are take i the thruster allcati algrithm. The sesitivity t the differet types f lsses depeds which type f prpeller ad thruster used, applicati f skegs ad fis, hull desig ad peratial philsphy (Leh, 992, Kaase, 996). Mai prpellers are subject t large thrust lsses due t air vetilati ad i-ad-ut-f-water effects. Rtatable azimuth thrusters are subject t dmiatig lsses caused by hull fricti ad iteracti with ther thrusters. Tuel thrusters are subject t lsses caused by -axial iflw due t curret ad vessel speed ad vetilati pheme i heavy weather. The effects f the abve metied lsses are mdelled by the thrust ad the trque reducti fuctis h ad h T based a cmbiati f aalytical ad empirical data. I ig. thrust
reducti fuctis (Leh, 992) accutig fr iad-ut-f-water effects are illustrated. Hece, the actual thrust a ad trque T a experieced by the prpeller ca be expressed as: = h (, x, θ) ( t ) = f (, x, θ), (2) a d T = h (, x, θ) T = f (, x, θ ), (3) a T T where x represets dyamic states, such as vessel mti, prpeller submergece ad evirmetal cditis. Prpeller depedet parameters are represeted by the vectr θ. 2.4 Prpeller Shaft Mdel Let T m dete the trque geerated by the thruster mtr. A trque balace (ig. 2) fr the prpeller shaft is writte: J ω = T T, (4) m where J is the mmet f iertia fr the shaft, ad ω = 2 π is the agular shaft speed. a trque ctrller may fr practical reass be assumed t be equivalet with: Tm ( s) = T s ( s ), (9) + where the time cstat T T is i the rage f 20-200 ms. Ψc alc - Ψcalc - Trque ctrller lux ctrller T PWM Mtr Mdel ig. 3 Trque lp i electrical mtr drive. 3. Speed Ctrl 3. CONTROLLER DESIGN I sa I sb m ~ ~ Iducti Mtr T m T a f T 2π Js f ig. 2 Prpeller shaft mdel (s detes the differetial peratr). The pwer delivered by the mtr ad the actual prpeller shaft pwer accutig fr the effect f thrust lsses are give by: P = ωt = 2 πt, (5) m m m P = ωt = 2 πt. (6) a a a By cmbiig equatis (), (2) ad (6), it is pssible t express the prpeller thrust as fucti f prpeller pwer ad trque accrdig t: / 3 2/ 3 ρ D K Th ( t d ) 2/ 3 a = sg( ) P / a, (7) 2 3 2π K h ( Q T ) a K T = K D T a. (8) 2.5 Trque Lp i a Electrical Mtr Drive Q The trque ctrl is iheret i the desig f mst applied ctrl schemes fr variable speed drive systems. The trque is ctrlled by meas f mtr currets ad mtr fluxes with high accuracy ad badwidth, see ig. 3. Theretically the rise time f the trque i PWM (Pulse Width Mdulated) drives is limited by the mtr s iductace (i lad cmmutated iverters, LCI s, als by the DC chke). Hwever, i practice the ctrller limits the rate f chage f trque i rder t prevet damages the mechaics. The a I cvetial PP systems a speed ctrller is used t achieve the cmmaded prpeller frce, as shw i ig. 4. The referece (cmmaded) speed i the prpeller ref is fud by the statiary fucti: ref ref 0 ref ref 4 ρd K T0 = g ( ) = sg( ), (20) which is the iverse fucti f the mial shaft speed-versus-frce characteristics give i () fr typically K = K ( J = ). g 0 T0 T 0 ref T a f T 2π Js PID ig. 4 Speed ctrlled prpeller. +T T s f The rated (mial) trque ad pwer are deted T N ad P N. The speed ctrller is a PID ctrller with saturati limits, which calculates the ecessary trque t btai icreased/decreased speed. The maximum trque is: T max = α T N, (2) where α is typically i the rage f. -.2. 3.2 Trque Ctrl I the prpsed trque ctrl strategy the uter speed ctrl lp is remved, ad the thruster is T m a
ctrlled by its trque ctrl lp with a cmmaded trque as set-pit, ig. 5. is calculated frm the referece trque T ref, which is derived by the statiary fucti: Tref = gt0( ref ). (22) Let the trque referece be writte: Tref = ρd 5 K Q0 ref ref (23) where typically K = K ( J = ). By cmbiig Q0 Q 0 (20) ad (23), the mappig betwee the cmmaded thrust frce ad the trque becmes: DK Q0 Tref = ref = g T0( ref ). (24) K T0 ratigs ly, but als by the available pwer i the geeratrs, this limit will vary accrdigly. By this methd the pwer limitati will becme fast ad accurate, allwig t utilise the pwer system s pwer capability with a built-i black ut preveti. T Max T N Reduced P Max P N P max 2 π P max, T max N g T0 T ref Trque algrithm +T T s T m ig. 7 Trque limitati. 3.3 Pwer Ctrl T a f T 2π Js ig. 5 Trque ctrlled prpeller. f T the existig trque lp a ew fucti, Trque algrithm, is added (ig. 6). is the cmmaded prpeller frce as befre. The trque referece is filtered thrugh a referece geeratr, which yields smth desired trque refereces T d ad T d. I rder t speed up the respse a referece feedfrward ctrl acti is cmputed. The cmmaded trque must als be limited by the maximum trque capability T max ad pwer capability P max. T ref Referece geeratr T d Referece feedfwd. T d a A alterative ctrl strategy fr prpeller ctrl based pwer ctrl is als ivestigated. The trque ctrl lp is maitaied, but the cmmaded trque is fud frm a cmmaded pwer P ref (ig. 8). This pwer referece is a siged value i rder t determie the trque directi. By cmbiig (20) ad (23), the mappig betwee the referece thrust frce ad the pwer becmes: 2πK Q0 3/ 2 Pref = sg( ref ) ref = g P ( ref ). (25) 3/ 2 ρdk T0 T the existig trque lp a ew fucti, Pwer algrithm, is added (ig. 9). The pwer referece is filtered thrugh a referece geeratr, which yields smth desired trque refereces P d ad P d. I rder t speed up the respse a referece feedfrward ctrl acti is cmputed. The cmmaded pwer is limited by the maximum pwer capability P max ad the maximum trque capability as befre. g P P ref Pwer algrithm P max, T max P c ( ) 2π +T T s T m mi Pmax Tmax, 2π T a f T ig. 6 Cmmaded trque algrithm. The trque limitig fucti is described i ig. 7. The trque must als be limited by the maximum pwer, which yields hyperblic limit curves fr the trque as a fucti f speed. Sice maximum pwer is t limited by the cverter ad mtr 2π Js ig. 8 Pwer ctrlled prpeller. f a
P ref Referece geeratr P d Referece feedfwd. P d { 2π Tmax Pmax} mi, ig. 9 Cmmaded pwer algrithm. P c the same parameters as used i ig. fr P/D = 0.89, the sesitivity fuctis s, s T, ad s P (26-28) ca be cmputed as a fucti f advace rati J, see ig. 0. As expected all three sesitivity fuctis decrease, resultig i icreased thrust lsses, fr icreasig J. The speed ctrl scheme is less rbust fr variatis i J, while the trque ctrl scheme shws the best rbustess. 3.4 Sesitivity t Thrust Lsses 0.8 I rder t cmpare the differet ctrl schemes a simplified sesitivity aalysis is carried ut t address the mst sigificat prperties f speed ctrl, trque ctrl ad pwer ctrl. Pitch ctrl is t studied, but is expected i geeral t have the same behaviur as speed ctrl 3.4. Speed Ctrl Assume that the shaft speed equals the speed referece ref at steady state (ig. 4). Usig (), (2) ad (20), the relatiship betwee actual prpeller thrust ad the referece thrust becmes: K T a = h ( )( t d ) K = s ( ) (26) 3.4.2 Trque Ctrl T0 rm a similar csiderati f the trque ctrl scheme (ig. 5), it is assumed that the mtr trque ad the prpeller trque equal the trque referece T ref at steady state. Hece, cmbiig (), (2), (2), (3), (20) ad (23) yields: h t K ( )( d ) Q0 K T a = h K K = s T( ) (27) ( ) 3.4.3 Pwer Ctrl T Q T0 r the pwer ctrl scheme (ig. 8) the prpeller ad mtr pwer ted t the pwer referece P ref at steady state. Hece, cmbiig (), (2), (2), (3), (6), (20), (23) ad (25) yields: a h = = s ( ) P ( )( t K d ) 2/ 3 h ( ) K T ref Q0 3.4.4 Psitiig Perfrmace Q 2/ 3 K T K (28) T0 The effect f psitiig perfrmace fr the differet ctrl schemes is clearly see frm the tw examples illustrated i ig. 0 ad ig. 2. Example. Let t d = 0 ad h = h T =. By usig (3), (4), K = K ( J = ) ad K = K ( J = ) with T0 T 0 Q0 Q 0 0.6 0.4 0.2 0 0 0.2 0.4 0.6 0.8 ig. 0 Sesitivity fuctis fr the differet ctrl schemes: s (slid), s T (dash-dt) ad s P (dash) as fucti f advace rati J fr P/D= 0.89. Example 2. Thrust lsses (ig. ) caused by iad-ut-f-water effects ca be mdelled by the thrust ad the trque reducti fuctis h ad h T (2-3) as a fucti f the submergece f the prpeller shaft h dived by the prpeller radius R. 0.8 0.6 0.4 0.2 0-0.5 0 0.5.5 2 ig. Mea thrust lsses h (dash-dt), h T (slid) ad reduced disk area (dash) due t reduced submergece i calm water f a heavily laded prpeller as a fucti h/r fr P/D= 0.89, based ig. 4. i Leh, 992. The reduced disc area due t the reduced submergece is accuted fr i the cmputati f (3) ad (4) accrdig t Leh, 992. Let t d =0, K = K ( J = ) ad K = K ( J = ) with the T0 T 0 Q0 Q 0 same parameters as used i ig. fr P/D = 0.89, the the sesitivity fuctis s, s T, ad s P (26-28) ca be cmputed as a fucti f h/r, see ig. 2. As expected all three sesitivity fuctis decrease, resultig i icreased thrust lsses, fr decreasig h/r. The speed ctrl scheme is less rbust fr reduced submergece, while the trque ctrl scheme shws the best rbustess.
0.8 0.6 0.4 0.2 0-0.5 0 0.5.5 2 ig. 2 Sesitivity fuctis fr the differet ctrl schemes: s (slid), s T (dash-dt) ad s P (dash) as fucti f h/r, h ad h T. 3.5 Electrical Pwer Plat Netwrk Perfrmace ad Stability The pwer plat ships cstitutes f several gas turbie r diesel egies drivig electrical geeratrs. r safe perati there must i all lad cditis be eugh spiig reserves with sufficiet available pwer fr upredicted variatis i lad i rder t prevet black-ut. The black-ut preveti by meas f reducig lad heavy csumers must typically respd faster tha 500ms t be effective. With trque ad pwer ctrl, the prpeller lad is less sesitive t variatis i the surrudigs, givig less pwer disturbaces the etwrk ad imprved vltage ad frequecy quality. Additially, the maximum pwer csumpti may easily be limited t the available pwer i bth schemes, sice the pwer limitati is explicit i the trque ad pwer ctrl algrithm. This i ctrast t speed ctrlled ad pitch ctrlled prpellers, where the actual pwer lad must be measured as a feedback sigal with a iheret time lag which deterirates the black-ut preveti respse time. The accurate ad fast ctrl f pwer ad pwer limitati i trque ad speed ctrl gives less upredicted lad chages, ad less eed fr available pwer. The umber f ruig geeratrs ca be reduced, ad the average ladig will be higher. This i terms gives less tear, wear, ad maiteace f the prime mvers. 4. SIMULATIONS Numerical simulatis are carried ut i rder t illustrate the prperties f the differet ctrl schemes with respect t the psitiig perfrmace ad the crrespdig effects the electrical pwer plat etwrk. Oly thrust lsses due t axial water iflw accrdig t (7) is csidered. It is assumed that the prpeller has the characteristics as i ig. with P/D = 0.89. The prpeller is expsed t axial curret equal t m/s ad waves described by a Piers-Mskwitz spectrum with sigificat wave height H s = 3.0 m, T p =0 s. A cstat prpeller frce referece (setpit) = 00 kn is cmmaded. I ig. 3 the effect prpeller speed (RPM) is shw. r the speed ctrl scheme the crrespdig speed setpit is equal t 00.9 RPM. As expected the pwer ad trque ctrl schemes will prduce higher speed i rder t cmpesate fr the lsses experieced. 05 04.5 04 03.5 03 02.5 02 0.5 0 00.5 0 0 20 30 40 50 ig. 3 Prpeller speed (RPM) applyig speed ctrl (lwest), pwer ctrl (middle) ad trque ctrl (upper). 40.5 40 39.5 39 38.5 38 37.5 37 36.5 36 0 0 20 30 40 50 ig. 4 Mtr trque (knm) fr speed ctrl (lwest), pwer ctrl (middle) ad trque ctrl (upper). I (ig. 4) the effect mtr trque is shw. r the trque ctrl scheme the crrespdig trque set-pit is equal t 40.0 knm. As expected the pwer ad trque ctrl schemes will prduce higher trque i rder t cmpesate fr the lsses experieced. 440 430 420 40 400 390 380 0 0 20 30 40 50 ig. 5 Mtr pwer (kw) fr speed ctrl (lwest), pwer ctrl (middle) ad trque ctrl (upper). I ig. 5 the effect mtr pwer is shw. r the pwer ctrl scheme the crrespdig pwer
set-pit is equal t 423 kw. It is clearly see that the pwer ctrl scheme results i lwest pwer peaks, while the speed ctrl results i highest pwer variace. I ig. 6 the actual prpeller frce prduced is shw applyig the speed ctrl scheme (lwest), the pwer ctrl scheme (middle) set ad trque ctrl (upper). The statiary ffset i actual prduced prpeller frce ca after a perid f time be cmpesated by the feedback acti i the psitiig ctrller. The time lag fr such a cmpesati results i reduced badwidth. 04 02 00 98 96 94 92 90 88 86 0 0 20 30 40 50 ig. 6 Actual prpeller frce (kn) fr speed ctrl (lwest - dt), pwer ctrl (middle - dash-dt) ad trque ctrl (upper - slid). 5. DISCUSSIONS Trque ctrl has bee shw t have the lwest sesitivity i btaied thrust with respect t disturbaces ad thrust lsses ad is thus the mst rbust ctrl strategy amg the evaluated alteratives. The variace i pwer is sme higher tha fr pwer ctrl, givig mre disturbaces i the pwer etwrk. Where pwer variati ad quality is essetial, the pwer ctrl may be a alterative. Hwever, the perfrmace at lw speeds ad thrust demad is expected t be pr sice small variatis i pwer referece gives large variatis i thrust. Hece, the algrithm will be highly sesitive t mdellig errrs. r such applicatis, a cmbiati f these tw strategies where trque ctrl is dmiatig at lw speed ad pwer ctrl at higher speed shuld be applied. Speed ctrl ad pitch ctrl are assumed t have similar characteristics with respect t sesitivity t disturbaces ad thruster lsses. 6. CONCLUSIONS A ew methd based mmet ad pwer ctrl f the prpeller ad thruster devices is prpsed. Istead f calculatig the speed ad pitch set-pit based a frce t speed/pitch relati, frce t mmet ad frce t pwer relatis are used. This methd gives a sigificat imprvemet i the perfrmace f the psitiig accuracy due t imprved rbustess with respect t thrust lsses. I additi the perfrmace ad the stability f the electrical pwer plat etwrk are imprved due t mre direct access ad ctrl f the actual pwer csumpti frm the thruster system. Simulatis illustrate the differece i perfrmace ad pwer plat stability f the differet ctrl schemes. ACKNOWLEDGEMENTS - ABB Idustri AS, Maritek AS ad The Research Cucil f Nrway are gratefully ackwledged fr fruitful c-perati ad ctributis. REERENCES Berge, S. P. ad T. I. sse (997). Rbust Ctrl Allcati f Overactuated Ships: Experimets with a Mdel Ship. IAC Cferece MCMC 97, Brijui, Cratia. Blake, M. (98). Ship Prpulsi Lsses Related t Autmatic Steerig ad Prime Mver Ctrl. Ph. D. Thesis, Servlabratry, Techical Uiversity f Demark, Lygby. sse, T. I. (994). Guidace ad Ctrl f Ocea Vehicles. Jh Wiley ad Ss Ltd. Grimble, M. J. ad M. A. Jhs (988). Optimal Ctrl ad Stchastic Estimati, Vl. (2). Jh Wiley & Ss. Healey, A. J., S. M. Rck, S. Cdy, D. Miles ad J. P. Brw (995). Tward a Imprved Uderstadig f Thrust Dyamics fr Uderwater Vehicles. IEEE Jural f Oceaic Egieerig, Vl. (20). N. 4, pp. 354-360. Kaase, K. E. (996). uel Optimizig System - Mathematical Mdel f Thrusters. Restricted Maritek Reprt MT596-0006, 53059.08.0. prduced fr ABB Idustri AS. Leh, E. (992). Practical Methds fr Estimati f Thrust Lsses. Maritek Reprt MT5A92-003, 53003.00.06. Newma, J. N. (977). Marie Hydrdyamics. MIT Press, Cambridge, Massachusetts. Osterveld, M. W. C. ad P. va Ossae (975). urther cmputer-aalyzed Data f the Wageige B-screw Series. It. Shipbuildig Prgress, Vl. (22), pp. 25-262. Sørdale, O. J. (996). Thruster Allcati: Sigularities ad ilterig. IAC Wrld Cgress, S, CA, Vl. (Q), pp. 369-374. Sørdale, O. J. (997). ull Scale Sea Trials with Optimal Thrust Allcati. IAC Cferece MCMC 97, Brijui, Cratia. Sørese, A. J., S. I. Sagatu ad T. I. sse (996). Desig f a Dyamic Psitiig System Usig Mdel-Based Ctrl. IAC J. f Ctrl Egieerig Practice, Vl. (4), N. 3, pp. 359-368. Walderhaug, H. (992). Mtstad, ramdrift, Styrig. Lecture Ntes, Departmet f Marie Hydrdyamics, Nrwegia Uiversity f Sciece ad Techlgy (i Nrwegia).