• Nie Znaleziono Wyników

A mathematical model for acceleration phase aerodynamically alleviated catamarans and minimizing the time needed to reach final speed

N/A
N/A
Protected

Academic year: 2021

Share "A mathematical model for acceleration phase aerodynamically alleviated catamarans and minimizing the time needed to reach final speed"

Copied!
12
0
0

Pełen tekst

(1)

J Mai- Sci Technol (2016) 2 1 : 4 5 8 ^ 7 0

D O I 10.1007/S00773-016-0368-Z CrossMark

O R I G I N A L A R T I C L E

A mathematical model for acceleration phase of aerodynamically

alleviated catamarans and minimizing the time needed to reach

final speed

M o r t a z a Azizi Yengejeh^ • H a n i i d Mehdigholi^ • M o h a m m a d Saeed Seif^

Received: 16 August 2015/Accepted: 9 January 2016/Published online: 16 February 2016 © J A S N A O E 2016

Abstract Racing catamarans use aerodynamic alleviadon concept which i n existing extreme ground effect signifi-candy enhances the performance. Beside design measures, controlling strategies may be employed as convenient solutions to improve the performance and address concerns regarding poor stability i n these crafts. B e i n g o f substantial importance f o r a racing catamaran to reach the final speed as soon as possible, this study attempts to find the optimal f o r m o f changing the drive angle (as control variable) to m i n i m i z e its acceleration time. I n this regard, a mathe-matical model is developed f o r f o r w a r d acceleration phase o f these catamarans based on empirical and theoretical methods. Then the formulation and solution algorithm f o r the time-optimal problem are described according to an indirect method. Results f o r a representative racing c r a f t have been presented i n uncontrolled and controlled con-ditions. Problem i n controlled condition has been solved w i t h o u t and w i t h a predefined constraint regarding stability margin. Optimal controlling o f the drive angle without stability constraint during the acceleration results i n 40 % reduction i n time required to reach the speed o f 110 k n and 14 % reduction i n resistance at this speed i n comparison to the uncontrolled case. A d d i t i o n o f the stability constraint changes optimal solution f o r drive angle and causes c r a f t t r i m angle f o l l o w a decreasing trend at higher speeds.

H Mohammad Saeed Seif seif@sharif.edu

' Center of Excellence in Dynamics and Hydrodynamics o f Marine Vehicles, Sharif University o f Technology, PO Box, Tehran 11365-9567, Iran

K e y w o r d s T i m e - o p t i m a l control • Racing catamarn • Aerodynamic alleviation • Extreme ground effect • M i n i m u m principle

1 I n t r o d u c t i o n

Aerodynamically alleviated catamaran concept has been used since more than a half century ago i n the f o r m o f tunnel hulls i n boat races. Superior performance o f these hybrid crafts i n comparison to conventional planing boats very soon led to introducing new separated racing classes [1]. These catamarans are indeed a subset o f the more general group k n o w n as aerodynamically alleviated marine vehicles ( A A M V ) . I n A A M V s aerodynamic and hydrody-namic l i f t s both are important i n cruise speed. I n catamaran configuration, hydrodynamic l i f t is generated by sponsons (Fig. 1) commonly i n the f o r m of asymmetric planing hulls. This asymmetry i n addition to g i v i n g stability i n turns [2] reduces wetted surface and removes the under h u l l spray drag [3, 4 ] . The structure between two sponsons is shaped c o m m o n l y i n the f o r m o f an a i r f o i l t o produce aerodynamic l i f t w i t h o u t any extra powered air support or skiiting. F r o m the aerodynamic viewpoint, this part acts as the ram w i n g or channel flow w i n g [ 1 ] . R a m - w i n g condi-tion happens when there is Httle or no gap between w i n g trailing edge and water surface, so a ( f u l l y or partially) stagnation flow occurs and nearly f u l l dynamic pressure w i l l exist under the w i n g . A t higher speeds w h e n w i n g sufficiently comes out f r o m water because o f presence o f side hulls, a restricted channel flow is f o r m e d under the w i n g . Aerodynamic configuration is very similar to the wings w i t h endplates.

As an eariy published works, Shipps [1] i n 1976 offered valuable discussions on different aspects o f ram w i n g

(2)

J Mar Sci Teciinol (2016) 21:458-470 459

Fig. 1 Aerodynamically alleviated racing catamaran

planing crafts. T w o years later, W a r d et al. [5] g i v i n g some technical data on the geometry, structure and propulsion system o f the racing catamaran K U D U I I , published results of some full-scale trial records on its smooth water per-formance. A t the same year, K a l l i o [6] performed com-parative trials between the K U D U I I and the conventional planing monohull, K A A M A . The measured data i n head sea state 2 showed that pitch m o t i o n f o r K U D U I I was about 40-70 % o f that f o r m o n o h u l l at speeds about 4 0 - 6 0 knots. Reif and Guenther [7] published the results o f w i n d tunnel and towing tank tests on model o f a tunnel racing boat. These results revealed the significant pordon o f aerodynamic drag i n total resistance o f the boat. Some years later, Reif [8] conducted another set o f model tests i n w i n d tunnel to study more accurately the aerodynamics o f a tunnel boat i n ground effect. I n another study i n these years, Nangia [9] reviewed some o f the design challenges of these crafts and made a f e w suggestions to address stability problems. One decade later i n 1997, Doctors [10] proposed a configuration called Ekranocat w i t h above mentioned principles and preformed a theoretical analysis to show its beneficial aspects. Some years later, M o r c h [11] by means o f experimental data and computational fluid dynamics (CFD) computations p r e f o r m e d an analysis on aerodynamic performance o f 7.5 m offshore racing cata-maran. M o r e recently, Russel [12] i n his book along w i t h reviewing the history o f tunnel boat design explains some design details o f tunnel boats w i t h a simple language. Recorded w o r l d speed records i n this reference f o r tunnel boats w i t h propeller propulsion systems and j e t engines are, respectively, about 180 and 275 knots. WiHiams [13] through his Ph.D. dissertation on aerodynamics o f high-speed multihuUed marine vehicles, using d i f f e r e n t analyt-ical and C F D analyses, tried to calculate the achievable superior peiformance f o r a planing catamaran as a result o f aerodynamic alleviation. Chancy and Matveev [14] used a mathematical model based on aerodynamic extreme-ground-effect theory [15] and hydrodynamic added-mass strip theory [16] to simulate the steady states and

vertical-plane motions o f a tunnel h u l l . Matveev [17] used the same theories f o r modeling the longitudinal motions o f a hydroplane boat i n w h i c h wing-like structure is extended behind the sponsons.

L i f t to drag ratio increases dramatically f o r a w i n g especially w i t h endplates i n extreme ground effect by decreasing the ground clearance. Rozhdestvensky [15] uses the'term "extreme ground e f f e c t " when this clearance lies below 10 % o f the cord. I n F i g . 2, Carter's experimental result [18] f o r m a x i m u m l i f t to drag ratio (which occurs at attack angle of 2 . 5 ° - 3 ° ) has been plotted against height to cord ratio w i t h and without endplates. W i n g aspect ratio is 1 and the height has been measured at traihng edge. A s can be observed at the most close proximity to the surface, i.e. hi c = 0.015, LID w i t h endplates is three times o f that at hi c — \. Endplates effect is noticeable f o r hic < 0.2 and at hi c — 0.015, L/D w i t h endplates is about 65 % more than that without them. However, such a gain i n l i f t is accompanied by g r o w t h i n aerodynamic pitching moment w h i c h makes the boat more susceptible to longitudinal instability. Indeed, at the h i g h speeds i n waterborne condition, there is a fine and f r a g i l e balance between the aerodynamic, hydrody-namic and propulsive forces. As a matter o f fact, the small aspect ratio wings used i n racing boats, on their o w n , are unstable i n the v i c i n i t y o f the surface [19]. Such critical nature o f the stability and safety concerns are the main reasons f o r the l i m i t e d fields o f application f o r these crafts.

Stability analysis o f A A M V s has been the subject o f some studies i n recent years. CoUu [20] provided a math-ematical f r a m e w o r k f o r the dynamics o f generic A A M V s and proposed a static stabihty criterion f o r them i n a sim-ilar f o r m to that o f I r o d o v [ 2 1 ] . Irodov static stability cri-terion is f o r W I G s and states; the aerodynamic center i n height should be located upstream o f the aerodynamic center i n pitch. The criterion proposed by C o f l u is inde-pendent f r o m aerodynamic and hydrodynamic centers i n pitch and so seems to be unrealistic. Irodov stability

S 12 s

11

=• With End] = Without E jlatc >

i'

'—-=• With End] = Without E ndplat \ ' 1%

11

II

il

! "

: •-a !

!i

1 1 J 1 i 1 0.01 0.1 1 h/c

Fig. 2 M a x i m u m l i f t to drag ratio versus the height to cord ratio at trailing edge o f a w i n g w i t h aspect ratio of 1 [18]

(3)

460 J Mar Sci Teclinol (2016) 2 1 : 4 5 8 ^ 7 0

criterion may be used directly f o r A A M V s similarly as that in Matveev and Kornev [19] by using aero-hydrodynamic centers i n height and pitch. I t means derivatives o f sum o f aerodynamic and hydrodynamic l i f t force and pitching moment should be used i n calculation o f centers i n height and pitch.

Kornev et al. [22] developed a dynamical model f o r racing boats based on combination o f theories developed for planing boats and W I G crafts. They preformed a linear stability analysis and studied the influence o f different geometric and mass parameters on dynamic stability bor-ders. I n another related study, Gu et al. [23], along w i t h p e r f o r m i n g a C F D analysis on aerodynamic aspects and using a simple mathematical model, discussed the initial development o f pitch stabilizing systems f o r a 21-foot-long planing catamaran.

As another design challenge, A A M V s may o f f e r poor performances i n inidal acceleration mode because these vehicles are designed to operate at very high speeds where there is a very little draft and aerodynamic l i f t force con-tribution is sufficiendy increased. H i g h value o f the hydrodynamic drag i n acceleration phase, w h i c h manifests itself as the resistance hump, may stop speed-up phase and prevent i t f r o m reaching final cruising speed. Such a problem exists more seriously f o r W I G crafts w h i c h are designed to operate i n airborne conditions and should successfully pass the take-off phase and even f o r conven-tional planing crafts during transition f r o m displacement into planing mode. Besides design measures, f o r example, i n the power o f propulsion system to cope this challenge, craft attitude control mechanisms may be employed i n acceleration mode. Trimmable drives i n propulsion system, control surfaces such as t r i m tabs and approaches such as throttling the engine power are used c o m m o n l y i n planing vessels f o r this puipose. T r i m m i n g the drive angle may be the most c o m m o n and feasible solution. I n the case o f using such controlhng mechanisms f o r a racing boat, e f f i -cient and o p t i m u m control policies should be applied to cause the craft not only easily but also rapidly pass acceleration phase. I n this regard, the present article tries o f f e r a solution to this question as to how the thrust angle should be changed i n a trimmable drive system, i n order to craft pass the acceleration phase and reach a final speed i n the m i n i m u m possible time. This is a time-optimal control problem w i t h the drive angle as its control variable. Opti-mal control theory has been used previously f o r the m o t i o n control o f marine vessels i n the applications such as sea-keeping and maneuvering. Examples o f application o f optimal control f o r high-speed crafts can be f o u n d i n [24¬ 31]. I n the present study, authors use the theory introduced in [32] to solve above time-optimal control problem f o r an aerodynamically alleviated racing catamaran without and w i t h additional constraint regarding stability margin.

The first step i n dealing w i t h the problem is to define its plant or dynamic model on which the control process w i l l be applied. I n the present problem, this plant is the model of vertical plane m o t i o n o f the racing catamaran during f o r w a r d acceleration i n respect o f the time. I n order to develop this 3-degree-of-freedom ( D O F ) model, hydrody-namic, aerodynamic and propulsion system l i f t and drag forces and pitching moment should be calculated to be used in the equations o f motion. These equations then are written i n a state-space f o r m and comprise the dynamic model of the problem. For this puipose, empirical and theoretical methods are employed f o r calculating the hydrodynamic and aerodynamic forces and moments. For the propulsion system, we use a convenient model w h i c h includes the operational effect of the propeller, drive and engine. B y developing such dynamic model we w i l l be able to model the craft m o t i o n during acceleration time f o r any defined variations i n drive angle as input. I n the present article after introducing this dynamic model its results f o r a racing boat i n an uncontrolled condition are presented. Then solution algorithm o f the time-optimal control prob-lem is described according to an " i n d i r e c t " method and applied f o r obtaining the optimal drive angle signal f o r this boat. The paper closes w i t h some conclusions and recommendations.

2 D y n a m i c m o d e l

As mentioned, dynamic model w i l l be based on empirical and theoretical methods f o r calculating the forces and moments. Most o f these methods are based o n the assumption o f constant f o r w a r d velocity o f vehicle or steady-state condition. Thus, although the model is time dependent, we neglect the unsteady effects o f the craft f o r w a r d acceleration on forces and moments. Another part of the unsteady effects is related to the aero-hydrodynamic forces and moments associated w i t h craft unsteady surge, heave and pitch motions. These terms, conventionally, are considered by added mass and damping coefficients i n equations o f motions. These coefficients have been used i n linear stability and seakeeping analysis o f planing crafts b y many authors. See f o r example [16, 33, 37]. Katayama and Ikeda [34] used these coefficients to carry out the time-domain simulation o f a personal watercraft i n f o r w a r d accelerated m o t i o n f r o m rest. They calculated hydrody-namic coefficients f r o m their measured data or theoretical methods based on a potential theory and f o u n d that the added mass forces contribution are small enough to be ignored. The calculation o f these coefficients f o r A A M V s w h i c h should be performed by taking into account both hydrodynamic and aerodynamic flows is beyond the a i m and scope o f this w o r k . I n the present study, equations o f

(4)

J Mar Sci Teclinol (2016) 2 1 : 4 5 8 ^ 7 0 461

motion are solved in a quasi-steady manner by neglecting tliese unsteady effects. B y quasi-steady, we mean craft speed changes f r o m time step to time step as a result o f f o r w a r d acceleration, but steady-state attitude at corre-sponding speed is used at each time step.

2.1 Hydrodynamic force and moment model

Using xyz coordinate system w i t h its origin fixed on c a l m water level ( F i g . 3), x^it) and , r 5 ( / ) are defined as time-dependent vertical position o f C G and t r i m angle o f h u l l , respectively. Then FIH and F^H, the hydrodynamic forces in X and z direction and F j / / , hydrodynamic pitch moment about center o f gravity (CG), w i l l be a f u n c t i o n o f c r a f t f o r w a r d speed ii(t), Xsit) and xsQ).

FiH =fiH{u,X3,Xi) / = 1,3,5 ( 1 )

For prismatic planing h u l l , Savitsky [35] gives empirical formulas w h i c h can be used to calculate l i f t force F^H and the longitudinal center o f pressure Lp (see F i g . 3). Savitsky method p r i m a r i l y has been introduced f o r monohulls but i t can be used f o r catamaran configuration by some m o d i f i -cations. L i u and Wang [36] introduced interference coef-ficient Ai i n Savitsky's equations f o r l i f t and center o f pressure f o r this purpose. For t w o l i m i t cases (i.e. f u l l interaction and no interaction between demihulls), A,-equals 1 and \ / 2 , respectively. A t zero spacing (i.e. A,- = 1) two asymmetric demihulls f o r m one symmetric h u l l and u n m o d i f i e d equations are acquired. The interference coef-ficient value between t w o above l i m i t cases depends highly upon the h u l l separation distance, speed and wetted length of the vessel. Savistky equations f o r Cto (the l i f t coefficient f o r zero deadrise) and Lp after introducing the interference coefficient A,- w i l l be modified as f o l l o w s :

CiO = (I8O.V5/71)(0.012;. "'YA,- + 0.0055A,-A2-Y(c2r)) (2) -1 •

Lp = bXl 0.75 5.21C,2r/(A,;0 -F2.39 (3)

where x^ is i n radian, b the effective beam (summation o f two demihulls' beam), X mean wetted length-beam ratio, /• separation ratio (the ratio o f b to overall beam o f the catamaran B) and C,, ul{gBf'^ speed coefficient (beam Froude number). I t should be noted that I and C„ are defined here based on overall beam. N o w , Fj^ can be calculated as f o l l o w s :

F^H = O.SCtppii^B'-. (4)

where Cip is l i f t coefficient f o r deadrise angle o f [i i n degrees and is calculated as f o l l o w s :

CLU = Q o - 0.0065^ C; .0.6 10 ' (5) where 1 is expressed w i t h respect to keel and chine wetted lengths Lf: and Lc according to E q . 6. L^ and Lc f o r pris-matic hull can be approximated by Eqs. 7 - 9 w i t h respect to

X3 and Xs [ 3 7 ] : X = 0.5{LK+LC)/B LK = LCG + VCG/tanA-5 -l-xs/sinxs LC=LK- 0.50 t a n / J / [ X 5 ( 1 + 2 n , a x / V , ) ] Z m a x / V , = - 2 X lO"^/?^ - 7 X IQ-^ - 10'^p • 0.575 (6) (7) (8) 5 (9) In Eq. 7 LCG and VCG are the longitudinal and vertical position o f C G , respectively, f r o m transom stern and keel Hne as shown i n F i g . 3. I n Eq. 8 ZmaJV, is a parameter

dependent on deadrise angle according to E q . 9 w h i c h has

Fig. 3 The used coordinate system, hydrodynamic forces on hull and some other geometric definitions

A

FfiCosx5=Fi,f{+ Fysinxs

Ffl sin,T5 = ( F 3 H+Fr s i n x j )tan X5

(5)

462 J M a r Sci Technol (2016) 2 1 : 4 5 8 ^ 7 0

been obtained by curve f i t t i n g on results o f a similarity solution presented i n [ 3 7 ] . Equations 6-9 are used f o r racing catamarans i n this study by neglecting interference and asymmetry effects on wetted area pattern. B y esti-mating the viscous drag, Fy, by an empirical f o r m u l a such as I T T C - 5 7 [37], FIH and F^H can be obtained according to Fig. 3 as f o l l o w s , hy is the moment arm o f Fy about C G .

F\H = -[Fv cos A-5 - I - {F-iH + Fy siuAs) tanA-5] (10) FsH = {Fm + Fv sin A's)/cos A-5 x {Lp - LCG) - Fyhy

( 1 1 )

Even though these formulas are v a l i d f o r extensive range o f beam Froude number, they can be used only when the bow is out o f water. The bow i m m e r s i o n may occur at l o w Froude numbers as a result o f l o w t r i m angle. W e assume that h u l l length is enough to prevent bow immersion.

2.2 Aerodynamic force and moment model

A l l o f the vessels experience aerodynamic force because o f the air flow on their unwetted parts. For the tunnel hull boats there is an additive portion due to the air flow w i t h i n the tunnel under the h u l l . These t w o sources o f aerody-namic force are treated i n d i f f e r e n t manners and are cal-culated separately i n the present study.

Force and moments due to the out o f water parts o f sponsons and i n general, upper structure, are estimated by convendonal empirical aerodynamic coefficients according to the craft's instantaneous attitude. On the other hand, force and moments due to the flow w i t h i n the tunnel are calculated by I D mathematical model o f w i n g w i t h end-plates i n the extreme ground e f f e c t introduced by Rozhdestvensky. For a w i n g w i t h endplates i n a close p r o x i m i t y o f surface when there is a small clearances under the tips o f the endplates, the flow nature is dominantly one-dimensional [ 1 5 ] . Simple nonlinear one-one-dimensional models has been used previously f o r example by Gafling-ton et al. [38] f o r steady flow under a w i n g w i t h endplates or by T u c k [39] f o r unsteady flow under 2 D w i n g section i n the extreme ground effect. Gallington et al. [38] considered the effect o f leakage o f flow under the tips o f the endplates

but chordwise variations o f flow velocity (and so pressure) under the w i n g have not been taken into account. Rozhdestvensky [15], later, extended this model, by taking into account the chordwise variations and inserting unsteady effects. He also, by i n v o l v i n g the v e l o c i t y c o m -ponent due to the laterally leaking flow, represented an analogous 2 D model f o r the problem. These I D or 2 D models have been used frequendy by matveev et al. (see for example [14, 17, 40]) f o r different types o f A A M V s . 2D model is suggested i n cases w i t h significant side leak-age. The authors o f the present study utilized b o t h I D and 2D steady models f o r some cases including typical range o f operation o f tunnel boats and i n spite o f significantly more computational e f f o r t o f 2 D model, observed accept-able coincidence f o r l i f t and drag coefficients ( w i t h error approximately 10 % i n average) between results.

Similar to Matveev and Kornev [19] i n c a l m water we consider a flat tunnel ceiling and an undisturbed water surface. Effects o f deformable water were recognized to be small f o r wings w i t h no augmented airflow b y Matveev [40].

The coordinate axis and geometry is depicted in F i g . 4. The governing one-dimensional continuity equation f o r steady inviscid and incompressible airflow w i l l be as f o l l o w s :

; f [ / . . ( x ) « . ( A ) ] - H ^ s i g n ( t / , ^ - . ? ( A ) )

<i-^' (12) X .^Uj - 11^{x) = 0, Us{0) = -Us

where ii^ is local spanwise-averaged velocity and the span. Sign f u n c t i o n has been added to ensure correct flow direction i n gap under endplates. Span-wise pressure dis-tribution according to B e r n o u l l i equation w i l l be as f o l l o w i n g :

Ps{x) = ^^P„x[U^-i<M (13)

Subsequently, aerodynamic l i f t , drag and p i t c h moment due to the air flow i n the tunnel can be calculated b y integration o f pressure distribution along x and adding viscous and leading edge suction effects according to [ 1 5 ] .

(6)

J M a r Sci Teclinol (2016) 21:458-470 463

Fig. 5 Wetted lengths on flat plate

Total aerodynamic drag, l i f t and pitching moment about CG denoted, respecdvely, by FIA, F^^ and F j ^ w i l l be determined by adding upper structure and tunnel flow portions. Aerodynamic model i n general can be represented by f o l l o w i n g functions:

FiA =fiAill,X3,X5) / • = 1,3,5 (14)

It has to be noted when trailing edge is below the undis-turbed water level, as a result o f airflow blockage i n tunnel.

Us = 0 and stagnation pressure w i l l be exerted under the

dry part o f the w i n g . For wetted part, wetted length-beam rado can be calculated f r o m Eq. 15 [ 3 5 ] . Ave and Xwi are defined i n F i g . 5. Hydrodynamic force on flat plate (i.e. /? = 0) can be calculated according to the same relations w h i c h were used i n Sect. 2 . 1 :

, r i . 6 1 , , 1 - 0 . 3 4 , ( 0 < i „ i < l ) / , 5 ^ U „ a + 0 . 3 ( l < A i n < 4 ) ^ ^

2.3 F o r c e and moment model for propulsion system

Considered model f o r calculating the force and moment exerted by the propulsion system on h u l l is same as that used by the present authors i n [ 3 2 ] . The model gives pro-peller thrust by taking into account the propro-peller-engine operational interaction i n a quasi-steady manner and also most o f the concerned peiformance factors. A c c o r d i n g to this model, propeller r p m and thrust change during the c r a f t f o r w a r d acceleration. A t n m m a b l e sterndrive is used as drive system and hydrodynamic drag and l i f t on its immersed part are considered.

This model finafly can be represented by f o l l o w i n g functions ĥ/>:

NpFip =fip{ii,X3,X5,(Pp) r = 1,3,5, (16)

where Np is the number o f engine-drive sets used i n propulsion system. Fip and Fi,p are horizontal and vertical forces due to the propulsion system on h u l l and F^p is their moments about C G o f vessel, (pp which w i f l be considered as control variable i n next sections is the propeller shaft angle relative to keel line, positive when propeller goes d o w n .

2.4 Equations of motions and the solution method in time domain

As mentioned, h u l l heave and pitch velocity and acceleration effects on fluid forces are neglected here. I n the other words, added mass and damping coefficients w i l l not exist i n equations. Only added mass i n advance direction o f craft due to fhe acceleration i n this direction, i.e. w i l l be used i n equations o f motion. As mentioned before, h u l l m o t i o n w i l l be considered accelerated only i n f o r w a r d direction so f o r vertical forces and pitch moments only e q u i l i b r i u m condi-tion w i l l be considered at each time step. Such assumpcondi-tion is true i n c a l m water because o f small rate o f change i n craft attitude i n these directions relative to change i n x-coordinate. The governing equations i n respect to introduced f o r c e models are as Eqs. 17-19. These equations should be solved simultaneously f o r 3 unknowns, ii, x^ and ^ 5 . (pp is k n o w n or w i l l be determined f r o m o p t i m a l control law.

{in + an)ii = a ' s . A s ) +fiAiii,X3,X5)

+ flp{ll,X3,X5,(pp)

= fi{ii,X3,xs,(pp) (17)

0 =f3H{u,X3,X5) +f3A(ll,X3,X5) +f3p{ll,X3,X5,(pp) - mg

= f3{u,X3,X5,(pp)

(18) 0 =f5Hiu,X3,X5) +f5Aill,X3,X5) + fip{u,X3,Xf,, (pp)

= fs{ll,X3,X5,(pp) (19)

I n order to solve these equations i n each time step we use simply a f o r w a r d finite difference f o r m o f Eq. 17 as

i>n + an)'-^^^^=AH{'r,xrA^')

+frAiir,xr,xr)

+f,p{u'+\x'+\x'+\(pp) (20) I n this equation superscripts indicate time step number and

«' is k n o w n f r o m i n i t i a l condition or previous t i m e step. Replacing x'3'^ and xs^' (unknowns at present time step) i n Eqs. 18 and 19 beside Eq. 20, we reach a system o f three nonlinear (algebraic) equations w h i c h can be solved by a numerical routine. Such procedure w i l l be continued at next t i m e step.

2.5 Modeling the acceleration phase

of an aerodynamically alleviated c a t a m a r a n without control

I n this section results o f introduced mathematical model are presented f o r a racing boat w i t h specifications very similar to K U D U I I [ 5 ] . The main characteristics o f the c r a f t are presented i n Table 1. Sponsons have been

(7)

464 J Mar Sci Technol (2016) 21:458-470

Table 1 Main characteristics of K U D U I I craft [5]

Sum o f demihulls' beam 2 X 0.96 m

Overall length 10.36 m

LCG 2.29 m

V C G 0.85 m

Mass 5361.6 kg

Deadrise angle 13°

Cord o f wing section 7.77 m

Span o f wing section 2.27 m

ho (Fig. 4) 0.32 m

a (Fig. 4)

considered as single-deadrise prismatic Irulls w i t l i neglec-ted inteiference effect. Before modeling the acceleration phase some comparisons between ward et al.'s results [5] and present model f o r attitude of the vessel i n steady-state condition have been presented i n Table 2. Measurements have been performed during fuU-scale trials i n three con-stant speeds. The c r a f t attitude has been calculated by present model o f hydrodynamic and aerodynamic forces using the k n o w n propulsion forces at these speeds. W h i l e general agreement between calculated and measured values is encouraging, significant discrepancy i n some points can be observed w h i c h is not so unexpected due to the sim-plifications i n model. It has to be noted that part o f the discrepancies are attributed to measurement errors as a result o f difficulties i n record o f t r i m angle and keel wetted length as mentioned by ward et al. [ 5 ] . However, there is a very little published data on performance o f these cata-marans and p e r f o r m i n g more accurate comparisons needs further experimental data.

I n this part we use the introduced dynamic model to sim-ulate the acceleration phase o f this racing boat. Solution algorithm i n time domain is according to the Sect. 2.4 and time step is 0.05 s w h i c h is sufficiently small to prevent f r o m time-step-dependent results. A t this stage, no control is per-f o r m e d on drive angle and it is per-fixed on - 9 . 5 ° . The same angle has been used f o r steady-state measurements by ward et al. [5]. A more p o w e r f u l propulsion system i n comparison to that f o r K U D U I I is considered i n order that the craft be able to reach speeds more than 100 k n . A couple o f sterndrive- engine systems, each w i t h m a x i m u m power of 1100 hp and the gearbox reduction ratio o f 1.25, are used i n propulsion system. The engine f u l l throttle power versus its i p m and open water performance curves o f propeller including thrust coefficient KT, torque coefficient Kg and efficiency ;/ versus advance ratio J are shown i n F i g . 6.

I n F i g . 7a craft speed variation has been plotted during i n i t i a l 30 s o f acceleration phase. A f t e r 22.6 s speed reaches 110 k n and finally steady-state condition has been achieved.

Table 2 Comparisons between the results of the of Ref. [5] and present model f o r K U D U I I craft

Speed (knot) Ref. T r i m angle do

61 Present 4.1 2.5 0.14 Ref. [5] 3.7 3.41 0.1 69 Present 3.9 2.2 0.17 i Ref. [5] 4.6 2.13 0.15 78 Present 4.2 1.7 0.19 Ref. [5] 5.1 1.37 0.2 \ ^ • lOKQ e O O 0 O Eff. ( c ' ' O

{ 1 r 0.5 1 1.5 2

J

(b) 900 - 1 •• 1 1 800 -• i I - - - - ¬ 700 T ; ~ 600 -! i ' ^ ~ ~ [ 500 -i - Z f i ^ f \ I I 400 T ^1 r 300 ---^„0^\ 1 1-200 - T ' 1 100

\ 1 1

0-J i 1 i 2000 3000 4000 5000 6000 R P M

Fig. 6 Open-water performance curves o f the used propeller and f u l l thi-ottle power generated by engine at each rpm

I n Fig. 7b the height o f the trailing edge o f the w i n g and craft C G f r o m undisturbed water level have been plotted against the time. Negative values f o r trailing edge height u n t i l f = 3.6 s mean i t is below the undisturbed water level and the stagnation occurs f o r air flow i n the tunnel. W e c a l l this situation as wetted-wing condition. I n order that traibng edge can emerge f r o m water, a 0.5-m rise o f C G is needed w h i c h takes place after speed o f 40 k n as a result o f craft planing. As mentioned before hydrodynamic l i f t is imposed on wetted part o f the w i n g .

Variations o f the w i n g wetted length, besides hull keel and chine wetted lengths, have been shown i n F i g . 7c. B y

(8)

J Mar Sci Teclinol (2016) 2 1 : 4 5 8 ^ 7 0 465 (b) 10 15 20 time(s) 25 30 ( C ) i ' ' K e e l Wetted Length ' ' K e e l Wetted Length

I A i = " Chine Wetted Length • • • W i n e Wetted Lenath

1 S 1 i > 1 ^ A l : • i\ ' ^ i L l • ' ^ \ '. '"^; ^>». — — '. '"^; ^>». — — \ J j ; ^ (e) 0.; 10 15 20 time (s) 25 30 © 1

-<

0.2 0.1 0 10 15 20 time (s) 25 3D 1,2 1 0.8 0.6 ' 0.4 0.2 0 - 0.2 - 0.4 - 0.6 0

= " Trailing Edge Heiglit — — C G Height 15 20 time (s)

id)

<^ 'ail a < g •C 10 15 20 time (s) 25 (f) 15 20 time (s) 25 \ \ \ \

4

\

. 1 .

\ / \

\---Total Resistance — — Acceleration \ \ \ '• \ \ S 30 - 7 • 6 • 5 - 4 • 3 - 2 1 0 30

Fig. 7 Variations o f some performance characteristics o f the racing boat during forward acceleration time without the control of drive angle

Starting the channel flow under the tunnel, as a result o f different pressure distribution pattern under the w i n g rel-ative to the stagnation condition, the slope o f decreasing trend i n keel and chine wetted lengths changes signifi-cantly. D r y chine occurs after / — 6.75 s or u — 75 k n . Keel wetted length retains the decreasing trend u n t i l reaching final steady-state value o f 0.62 m . It has to be noticed that although the Savitsky's empirical relations are k n o w n to be used f o r wetted chine planing, they are used here also f o r dry chine condition by taking into account the change i n wetted beam o f planing surfaces. A c c o r d i n g to Payne analysis [41] such assumption is not so unrealistic f o r prismatic huUs w i t h deadrise angles between 10° and 15° i n t r i m angles between 4 ° and 10°.

Boat t r i m angle variations during acceleration are shown in F i g . 7d. T r i m angle experiences a hump and h o l l o w i n

wetted-wing condition. I t experiences second m a x i m u m and m i n i m u m , respectively, where w i n g t r a i l i n g edge clears the water and dry-chine condition occurs. Then i t foUows an increasing trend until reaching steady-state value.

The ratio o f aerodynamic l i f t to craft weigh during acceleration has been shown i n F i g . 7e. A t the final steady-state condition (ii = 110.5 k n ) aerodynamic l i f t contribu-tion reaches about 43 % o f weight. The remaining part o f weight is supported by sum o f the hydrodynamic l i f t and the vertical component o f the thrust.

Resistance and horizontal component o f the thrust dur-ing acceleration time have been plotted i n F i g . 7 f . I n this figure f o r w a r d acceleration also has been shown w i t h secondary vertical axis at the right-hand side. A t the beginning of the m o t i o n , the difference between thrust and

(9)

466 J Mar Sci Technol (2016) 21:458-470

resistance is significantly high and reladvely high values o f acceleration w i t h a m a x i m u m o f 6.5 m/s^ can be observed. The local m a x i m u m i n resistance at initial seconds is the convendonal resistance hump i n planing crafts which appears markedly small beside final resistance that resulted f r o m the high attainable speeds.

3 T i m e - o p t i m a l c o n t r o l p r o b l e m

As mentioned i n the introduction, the aim o f this study is to solve the optimal control problem o f m i n i m i z i n g the dme needed to reach a final speed f o r a racing catamaran. I n time-optimal control problems, the m a i n objective is find-ing control signals as input f o r a plant i n such a way that t i m e needed to go f r o m an initial to a final state be m i n i -m u -m . Also at the sa-me d-me, so-me constraints -must be satisfied. Constraints may be on control or state variables. I n the present problem, physical constraint on control variable, i.e. drive angle, is (pp^min < Vpi'^) — Vp.mm- Plant of the problem, which was expressed eventually by Eqs. 17-19, can be represented i n a more compact f o r m as:

[u{t) 0 0]^ =f{X{t),(pp) = \/{m + ax^WM.X3,Xi,q)p)

h{u,X-i,Xi,<Pp) fs{u,Xn,,X5, (pp)^

= \fl {ll,X3, Xs, (pp) ƒ 3 ( M , A-3, A-5, (pp)

fs{ll,X3,Xs,(Pp)]^ (21)

where superscript T stands f o r transpose, X ( 0 = [u(t)

A-3(f) xsit)f is the state vector, and X(t) and (ppit) are unknowns that should be determined i n time interval to to t f . The above equations construct a differential-algebraic system o f equations.

3.1 Solution algorithm for time-optimal control problem

T o solve the abovementioned problem different numerical approaches may be used i n the optimal control theory. W e use i n present study an indirect method w h i c h is based on M i n i m u m Principle. A l t h o u g h f r o m mathematical view-point this principle offers the necessary condition and on its o w n does not guarantee the globality o f the optimal solu-tion, it has been used since decades, i n solving many dif-ferent problems i n engineering applications.

I n indirect methods the problem is converted to a boundary value problem w i t h state and costate equations as its governing equations. State equations are indeed the plant equations and costate equations are added to find costate or adjoint variables.

Hamiltonian function f o r this problem is defined as f o l l o w s :

H{X, X,(pp) = l+ J2 ¥i(u,X3,X5,<Pp), '=1,3,5

(22)

where I = [li A3 Xsf is the vector o f costate variables. According to M i n i m u m Principle, as a necessary con-dition f o r (p*p, it should be obtained i n such a way that above H a m i l t o n i a n (when considered as a f u n c t i o n o f cpp) b c m i n i m i z e d f o r physically admissible values o f (pp so at each time we have

^ A ; / ; ( « * , A - ; , A - ; , r p ; ) < ^ ; . ; ƒ • ( « * ,A - ; , A - ; , rp^) (23)

'•=1,3,5 (=1,3,5

Star sign * denotes the optimal values o f variables.The state equations are obtained according to the dynamic model equations on the optimal trajectory as f o l l o w s :

f,{it*,x*,xi,(p*p)-ir =0 (24)

Mu*,xl,x*s,cp'p)=0 (25)

fsiit*,x*„x*s,cp;)=0 (26)

The costate equations w h i c h are obtained f r o m H a m i l t o -nian derivative w i t h respect to state variables on optimal trajectory, w i l l be i n the f o l l o w i n g f o r m : J2 KMit*,xl,x*s,(p*p) + X\=0 '•=1,3,5 i*fi,,{i,*,x;,x*s,cp;) = Q 1=1,3,5 ^ X%,{u*,xl,x*s,tp*p)=0 '•=1,3,5 W i t h Initial condition at t

It* Xj X*sY=^ [llo Xo3 •

fo as i T (Pp\, = <PPO (27) (28) (29) (30)

A t t = t f , craft should reach final speed and H a m i l t o n i a n must vanish as final condition:

1 + '•=1,3,5 X*Mit*,xl,x*,(p*p) 1 + r « * = 0, II \, = u„ i + xiMtt*,x;,x*s,cp*p) (31)

I n Eq. 27 fi„[ii*, xl,x*s, (p*p) is the partial derivative o f f u n c t i o n / i w i t h respect to u at the point (ii*,xl,x*s, (p*p). /ivs and fixs i n Eqs. 28 and 29 are also derivatives w i t h respect to X3 and A5 at this point. Above boundary value problem is solved here by a single shooting method. A t first time step, solution algorithm starts w i t h i n i t i a l con-dition and an i n i t i a l estimation f o r A j . Then A3 and A j are determined f r o m the solution o f Eqs. 28 and 29 w h i c h are linear algebraic when we use initial values. H a v i n g adjoint variables we can f o r m Hamiltonian as Eq. 22 and solve numerically, relevant constrained minimization problem to find (p*p at this time step. N o w we can solve state equations

(10)

J Mar Sci Teclinol (2016) 2 1 : 4 5 8 ^ 7 0 467

F i g . 8 The numerical solution algorithm of the time-optimal problem 'Costate Equations: =1,3,5 =1,3,5 '=1,3,5 A ^ O p t i m i z a t i o n P r o b l e m Minimize: / / ( X , ) . , ^ p ) = l + =1 ,3,5 Dynamic M o d e l (Pp State Equations: X 3 , X 5 , ^ ? ^ ) - Z / * = 0 y ^ ( i / * , x 3 , x 5 , ^ p ; ) = o

to update ( P j X j j X j . Finally, E q . 27 w i t h updated values is used to f i n d ?*•^ f o r next dme step. This procedure should be repeated at each time step u n t i l reaching the final condition, i.e. u* = Hma,x- A t this point i f the other f i n a l condidon o f E q . 31 was not satisfied by a defined error this numerical procedure should be repeated entirely, w i t h a new i n i t i a l guess (shooting) f o r X\. The numerical solution algorithm has been illustrated i n F i g . 8. A t the end o f the solution procedure, the final time t f , optimal

control signals (p*p{t) and optimum trajectory i n time domain, i.e. u*{t),xl{t) a n d x 5 ( ? ) are determined.

3.2 Solution of time-optimal control problem for K U D U I I without state constraint

I n Sect. 2.5 results o f the mathemadcal model f o r accel-eration phase i n a fixed tpp were presented. I n this secdon we apply the optimal control algorithm to find the (pp{t)

(11)

468 J Mar Sci Technol (2016) 21:458-470 ^\ \ /' I 1 ! i T -30 25 20 15 10 5 10 12 14 time (s)

F i g . 9 The optimal solution f o r drive angle and variations of the craft total resistance and trim angle during the control

w i t h w h i c h the same craft reaches the final speed of 110 k n i n the m i n i m u m possible time. Physical constraint on (pp is — 10° < <PP < 5 ° and at this stage no constraints on state variables are considered.

A s can be observed, f o r example, f r o m F i g . 7d, the state variables such as t r i m angle are not differentiable at points where the w i n g trailing edge clears the water and chines become dry. Thus to be able o f using above mentioned optimal control algorithm, according to the optimality principle, problem can be broken into three sub-problems i n w h i c h state variables are continuously differentiable. Indeed, the same procedure as F i g . 8 should be utilized f o r each sub-problem. However tbe boundary condidons f o r these sub-problems must be matched correcdy. I t means the i n i d a l condition of second sub-problem should be set same as the final state o f first sub-problem and so on f o r third sub-problem.

I n Fig. 9 the solution f o r (p), it) has been shown. I n order to facilitate tracing the dependencies on control variable, variations o f the craft total resistance and t r i m angle also are shown i n this figure. The resistance value should be read f r o m the axis at the right-hand side. The solution intervals f o r sub-problems have been indicated by vertical Unes. According to the obtained optimal solution, to m i n i m i z e acceleration time, drive angle should be adjusted at upper l i m i t o f - F 5 ° since r = 1.6 s (where t r i m angle experiences first local m i n i m u m ) then be lowered as i n F i g . 9 since reaching final value of about - 3 ° . A t the boundaries o f the separated intervals f o r sub-problems slight jumps (about 0.15° and 0.25°) can be observed as a resuk o f discontinuity i n derivatives. B y optimally controlling the drive angle, the boat reaches the final speed o f 110 k n i n 14 s. A t this speed t r i m angle and resistance are, respectively, 4 . 8 ° and 25 k N .

3.3 Solution of time-optimal control problem with additional constraint regarding stability

As mentioned i n the introduction, preserving the stability is one o f the main concerns i n design o f aerodynamically

time (s)

F i g . 10 The optimal solution f o r drive angle and variations o f the craft total resistance and trim angle during the control, w i t h additional constraint regarding stability

alleviated racing catamarans. Instability can be avoided by appropriate control strategies on craft t r i m angle. B y suf-ficiently l o w e r i n g the bow, craft skipper can prevent f r o m undesired pitch instability. Kornev et al. [22] using a hnear stabihty analysis f o r a racing catamaran presented stability border curves indicating the critical t r i m angles at w h i c h instability starts f o r each craft speed. B y such a manner, stable regions can be determined w i t h respect to speed and t r i m angle. I n other words, stability condition can be rep-resented by an inequality relation between u and xs- Such an inequality can be considered as a state constraint i n optimal control problem to avoid unstable planing. Con-cerning this, i n present section, the above t i m e - o p t i m a l problem is solved by additional constraint regarding sta-b i l i t y margin on state variasta-bles u and X5. Calculating the stability margins is out o f scope o f the present study so we assume i n this article a representative stability margin similar i n trend to those obtained by Kornev et al. [ 2 2 ] . There are d i f f e r e n t approaches such as the penalty f u n c t i o n method f o r solving the optimal control p r o b l e m w i t h state constraints. I n this method, inequality constraint is con-verted to equality constraint and the problem is solved by augmenting the H a m i l t o n i a n w i t h convenient additional term including this constraint.

I n F i g . 10 the obtained optimal solution f o r drive angle after implementing the additional constraint regarding stability has been shown along w i t h the craft t r i m angle and total resistance. Referring to F i g . 9 we can see that addi-tional constraint on state variables is active just after ? = 7.7 s where unconstrained t r i m angle curve violates the stability threshold. As can be observed f r o m F i g . 9, after this time, <ip*p stops the decreasing trend and starts to increase u n t i l reaching near the upper physical l i m i t o f -t-5°. This increasing trend o f thrust angle generates neg-ative pitch moment w h i c h tends to push d o w n the b o w . Optimal solution f o r drive angle is that w h i c h enforces t r i m angle to f o l l o w the predefined stabihty border curve. B y such trajectory f o r craft attitude, i t never reaches final

(12)

J Mar Sci Teclinol (2016) 21:458-470 469

Speed o f 110 k n because o f higher total resistance. So a slightly lower value equal to 108 kn has been considered as final condition. The craft reaches this final speed at 16.8 s and at this time, resistance value is 29.3 k N .

4 C o n c l u s i o n s a n d r e c o m m e n d a t i o n s

Aerodynamically assisted catamarans are kinds o f A A M V s w h i c h fill the gap between conventional planing and W I G crafts. The review performed i n this article on the pub-lished studies on these crafts indicates that there is a sig-nificant lack o f dedicated investigations and experimental data i n comparison to conventional planing and W I G crafts. A m o n g design challenges f o r these marine vehicles, their instability and poor performance during acceleration phase were emphasized i n this article. Beside primary design measures, controlling strategies were proposed as convenient solutions to counter these concerns. I n this regard, an optimal control problem f o r acceleration phase o f these crafts was intended i n which we seek f o r the optimal drive angle signal to m i n i m i z e the craft accelera-d o n time.

Results were presented i n uncontrolled and controlled conditions. Uncontrolled case was preformed w i t h fixed drive angle and simulation was continued until reaching steady-state condition. About 43 % o f weight o f the craft is supported by aerodynamic l i f t at final steady-state speed o f about 110 kn. The analysis prefoimed on relevant variables during the acceleration revealed a different behavior before and after when trailing edge o f the w i n g section clears the water.

The problem i n controlled condition was solved without and w i t h a predefined constraint regarding stability margin. Optimally controlling the drive angle without stabihty constraint during the acceleration resulted i n 40 % reduc-tion i n time required to reach the speed o f 110 k n and 14 % reduction i n resistance at this speed i n comparison to uncontrolled case w i t h fixed drive angle o f —9.5°. The addition o f the constraint regarding stability changes opti-m a l solution f o r drive angle i n such opti-manner that causes craft t r i m angle f o l l o w the stability border curve while the unconstraint problem violates the stability condition. This results i n higher resistances and slighdy lower final speed. A p p l y i n g this additional constraint increases the time required to reach final speed o f 108 k n about 33 % and resistance at this speed about 21 %.

As development o f work i n the future, more exact models f o r calculating hydrodynamic and aerodynamic force and moment may be used. Supplying more experi-mental data is necessary f o r validation purposes i n these models.

A more accurate dynamic model may be used by including aero-hydrodynamic added mass and damping

coefficients f o r solving the time-optimal problem i n rough water and more realistic sea condition i n further develop-ments. Optimal problem can be solved f o r other marine vehicles such as W I G crafts before take-off phase and w i t h other controlling mechanisms. This study can be consid-ered as i n i t i a l steps f o r designing and f u n c t i o n i n g the automatic close loop controllers f o r marine vehicles, w i t h the 'aim o f keeping the best performance at the same time preventing instability risks.

R e f e r e n c e s

1. Shipps PR (1976) H y b r i d ram-vving/planing craft—today's raceboats, tomorrow's outlook. In: A l A A / S N A M E Advanced Marine Vehicles Conference, no. A I A A 76877. Arlington, V i r -ginia, 20-22 Sept 1976

2. Bate J (1994) Performance analysis and prediction of high speed planing craft. Ph.D. thesis. University o f Plymouth

3. Yengejeh M A , A m i r i M M , Mehdigholi H , Seif M S , Yaakob O (2015) Numerical study on interference effects and wetted area pattern o f asymmetric planing catamarans. Proc Inst Mech Eng Part M J Eng M a r i t Environ. doi:10.1177/1475090215586638 4. X u L , Troesch A W (1999) A study on hydrodynamics o f

asym-metric planing surfaces. I n : Proceedings o f F A S T ' 9 9 , v o l 99, pp 471-481

5. Ward T M , Goelzer H F , Cook P M (1978) Design and perfor-mance o f the ram wing planing craft K U D U I I . I n : A I A A / S N A M E Advanced Marine Vehicles Conference

6. Kallio J A (1978) Results o f f u l l scale trials on two high speed planing craft (kudu I I and kaama). Tech. Rep. DTNSRDC/SPD-0847-01, D a v i d W . Taylor Naval Ship Research and Develop-ment Center

7. Reif T H , Geunther D A (1978) A comparative study o f the aerodynamics and hydrodynamics o f a tunnel hull boat. J H y -dronaut 12(4): 166-168

8. Reif T H (1985) A wind tunnel study o f the aerodynamics o f a tunnel boat h u l l w i t h consideration o f ground effect. High Speed Surf Craft 24(2):29-33

9. Nangia R K (1987) Aerodynamic and hydrodynamic aspects o f high speed water surface craft. Aeronaut J R Aeronaut Soc 91(906):241-268

10. Doctors L J (1997) Analysis o f the efficiency o f an eki-anocat: a very high speed catamaran with aerodynamic alleviation. I n : International Conference on W i n g in Ground E f f e c t Craft (WIGs 97), R I N A ed

11. M o r c h HJB (2003) Aerodynamic properties o f a high speed offshore racing catamaran. I n : Proceedings o f 7th International Conference on Fast Sea Transportation, Ischia, Italy, 7 - 1 0 Oct 2003

12. Russel J (2007) Secrets o f tunnel boat design. Aeromarine Press, Cambridge

13. W i l l i a m s A G W (2008) Aerodynamic forces on highspeed m u l -tihuUed marine vehicles. Ph.D. Thesis, Cranfield University 14. Chaney CS, Matveev K I (2014) Modeling o f steady motion and

vertical-plane dynamics o f a tunnel hull. Int J Naval Ar-chit Ocean Eng 6(2):323-332

15. Rozhdestvensky K V (2000) Aerodynamics o f a l i f t i n g system i n extreme ground effect. Springer Science & Business Media, Berlin

16. Martin M (1978) Theoretical determination o f porpoising insta-b i l i t y o f high-speed planing insta-boats. J Ship Res 2 2 ( l ) : 3 2 - 5 3

Cytaty

Powiązane dokumenty

A smooth weight is used to control error terms, and this weight can in typical applications be removed from the final result.. Similar results are obtained for the tails of

The preceding examples discourage hope that root systems are always the best way to construct pure products of small norm, but perhaps if we look at Table 3 comparing the best

The crucial step in the classification of Markov traces is the observation that the values of such a trace on basis elements of H ∞ corresponding to elements in signed block form in W

The first section of this paper deals with the control problem for our fourth-order operator, the second considers a second-order operator, and the third studies the

The formula for optimal feedback control, using a non-symmetric (in the general case) solution to the differ- ential operator Riccati equation non-solvable with respect to

But this paradigm also shifts when it comes to a different ambient, like in the case of fado, and I will now show how the literary meaning of Estado Novo may cause problems, again

We give some necessary and sufficient conditions for an n−1 times differen- tiable function to be a generalized convex function with respect to an unrestricted n- parameter

waar slavernij in “De familie Kegge” nog niet meer dan motief is, wordt het later thema in enkele redevoeringen van Nicolaas Beets, een in 1847 voor het Zeister