This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Tom 52 (66) Fascicola 3 ISSN 1224-6077

SCIENTIFIC BULLETIN

of the ”POLITEHNICA” University of Timişoara, Romania Transactions of Mechanics

**Proceedings of the 3nd Workshop on Vortex Dominated Flows Achievements and Open Problems
**

Timişoara, Romania, June 1-2, 2007

Edited by S. Bernad, S. Muntean, R. Susan-Resiga

Editor in chief Prof.dr.ing. Victor BĂLĂŞOIU

Editors Dr.ing. Sandor BERNAD, Dr.ing. Sebastian MUNTEAN, Prof.dr.ing. Romeo SUSAN-RESIGA

**Workshop Scientific Committee
**

Prof. R. RESIGA Charmain Dr. S. BERNAD Dr. S. MUNTEAN Prof. E.C. ISBĂŞOIU Prof. C. BERBENTE Prof. L. SANDU Prof. A. LUNGU Prof. L.I. VAIDA Prof. V. CÂMPIAN Dr. Th. POPESCU

Politehnica University of Timisoara

Romanian Academy – Timişoara Branch Romanian Academy – Timişoara Branch Politehnica University of Bucharest Politehnica University of Bucharest Technical University of Civil Engineering, Bucharest ”Dunărea de Jos” University Galaţi Technical University, Cluj-Napoca ˝Eftimie Murgu˝ University, Reşiţa ”Gheorghe Asachi” Technical University, Iaşi

Dan NIŢU Editorial adviser: Ştefan KILYENI Desktop publishing: Valentina ŞTEF Tiparul executat la Imprimeria MIRTON 300125 Timişoara. Str. Samuil Micu nr. 7 Telefon: 0256-272926.Editura ORIZONTURI UNIVERSITARE Typesetting: Digital data supplied by editors Cover-Design: Romeo SUSAN-RESIGA. Sandor BERNADA. 0256-225684 Fax: 0256-208924 .

2007. in Timişoara is aimed at evaluating the current developments as well as at correlating the scientific efforts among the parteners. naval propellers. The 20 papers reviewed by the scientific committee. and on developing new mathematical models and numerical techniques for both inviscid as well as unsteady turbulent swirling flows. 33 for the period 2005-2007. • Technical Civil Engineering University. Asachi” Iaşi. has organized its third Workshop on “Vortex Dominated Flows – Achievements and Open Problems” in Timişoara June 1 – 2. Bucharest. have been presented and discussed at the workshop then included in the present issue of the Scientific Bulletin of the “Politehnica” University of Timişoara. Although the ongoing grant ends in 2007. Aerospace Engineering School. Prof. • Romanian Academy – Timişoara Branch. Transactions on Mechanics. This research consortium is supported by the Romanian National University Research Council grant No. the Academic Consortium for Research and Development on Fluid Dynamics “ACCORD-Fluid” will continue its activity as a national research network. The 2007 workshop brings together eight research teams from leading universities and research institutes in Romania: • “Politehnica” University of Timişoara. • Technical University “Gh. June 1-2. wind engineering. combustion systems. Applications for turbomachines. • University “Eftimie Murgu” Reşiţa. • Politehnica University from Bucharest.FOREWORD The Academic Consortium for Research and Development on Fluid Dynamics (ACCORDFluiD). • Politehnica University from Bucharest.and hydro-dynamics. 2007. which brings together research teams from leading Romanian universities in a joint effort to address current challenges on vortex and swirling flows aero. Development of experimental techniques for vortex flows. Romeo SUSAN-RESIGA . The two-day meeting. in particular swirling flows in adverse pressure gradient.dr. and it is aimed at: Fundamental research.ing. Power Engineering School. • University “Dunărea de Jos” Galaţi. • Technical University Cluj-Napoca. focused on ellucidating the physics of vortex flows. Development of numerical techniques able to meet the special requirements for vortex flows.

.. Calin GHERGU………………………………………….................... Sebastian MUNTEAN.............. Radoslav NABERGOJ............................................. Numerical simulation and experimental results Adrian STUPARU.......................................................................................... A semianalitical method for the wing aerodynamics Marius STOIA-DJESKA.. Free-surface turbulent flow around a lPg ship hull Adrian LUNGU ...................................................... Andrei GEORGESCU.............................. Romeo SUSAN-RESIGA ……………………………………………………………………………….. Numerical analysis of pump hydrodynamics at constant speed Eugen Constantin ISBASOIU.. Marius STOIA-DJESKA...................... On the stick – slip boundary conditions at the wall of microchannels Corneliu BALAN................. Sanda-Carmen GEORGESCU. Simulation of the Ship Standard Manoeuvres Florin PACURARU...................... Dan OBREJA ..... Uncertainties in the free-surface potential flow code solution Ana-Maria TOCU........ Petrisor STANESCU............... Costin Ioan COŞOIU................................ Diana BROBOANA....... Swirling flows in the suction sumps....................... Marius STOIA-DJESKA ..................... Roland KADAR ....... Numerical and Experimental Investigation on a Tractor Tug Resistance Performance Corneliu BALAN... Development of a Swirling Flow Control Technique for Francis Turbines Operated at Partial Discharge Andrei-Mugur GEORGESCU.... Corneliu BERBENTE .................................. COMSOL MULTIPHYSICS Versus FLUENT: 2D Numerical simulation of the stationary flow around a blade of the Achard Turbine Sandor BERNAD. Sandita PACURARU ................. Lucian SANDU .... Daniel BALINT... Răzvan......................... Cavitation tip clearance........ Sterian DĂNĂILĂ..................................................... Diana BROBOANA ............................................... Sebastian MUNTEAN.Silviu ŞTEFAN.............. Dorian NEDELCU ... Pressure constrain in vicinity of the separation point in planar................... Liviu CRUDU.......... Costin Ioan COŞOIU ……………………………………………………………………………….............. Sandor BERNAD .............................. Andrei-Mugur GEORGESCU...... Sterian DANAILA......... Catalin BALAN............................ Performances data of propulsion Systems for high speed ships Dan OBREJA...................................... Liviu ANTON.............................TABEL OF CONTENTS Romeo SUSAN-RESIGA....... Liviu HAŞEGAN................... Carmen Anca SAFTA.......................................... Sandor BERNAD........... Danile BALINT. Adrian LUNGU ...... Carmen Sanda GEORGESCU. Some aspects about a vortex building model placed upwind an aeroelastic model in the boundary layer wind tunnel Adrian LUNGU …………………………………………………………………………………………........................... 2D unsteady simulation of the flow in the Achard marine turbine Corneliu BERBENTE. Numerical flow investigation over a chemical tanker hull Mihaela AMORARITEI ............ Experimental approach Mircea DEGERATU.......... Vlad HASMAŢUCHI...................................... A Physical And Theoretical Analysis of the Vortex breakdown on delta wings Constantin Viorel CÂMPIAN... Diana Maria BUCUR.................... Alexandru BAYA ........ Georgiana DUNCA.. steady and isochoric motion: case i–newtonian fluid ................

.............. Numerical investigation on a swirl burner with internal flue gas recirculation Victor HODOR................................ Corina GIURGEA .. Victor HODOR............... A foundation in distributions of Glauert theory ............................... Claudiu RATIU ………………………………........ Florin BODE........Florin BODE............................................................................ Paula UNGURESAN... CFD first prediction in designing a 50kw swirling burner within its combustion chamber Theodor POPESCU .................................

Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). Bucharest. wing. sail NOMENCLATURE An = coefficient in series expansion of circulation distribution λ = aspect ratio b = semi-wing span CD = induced drag coefficient CL = lift coefficient c = local wing chord y = spanwise coordinate on wing 1.1 The vortex system The lifting line model was essentially the replacement of the lifting wing by a model consisting of a system of vortices which imparted to the surrounding fluid a motion similar to the actual flow. The stagnation point moves towards the trailing edge. the trailing vortex system. there is a need for simplified models that allow for an easy grasp of physical dominant effects. Lifting line theory is still useful because it provides a relatively simple method for the analysis of the aerodynamics of different wing shapes and allows a relatively quick comparison of the performances of different wing geometries. and the fluid is unable to turn round the trailing edge to the stagnation point (on the upper surface). Instead it leaves the surface and produces a vortex just above the trailing edge. 011061. The theory may be expanded to provide estimates of the change in the lift created by flaps and the rolling moment coefficient produced by an aileron deflection. Department of Aerospace Sciences “E. 2. Prof.: (+40) 21 023967. Instead it decelerates gradually along the aerofoil surface.Carafoli” University Politehnica of Bucharest Marius Stoia-Djeska.Carafoli” University Politehnica of Bucharest *Corresponding author: Polizu 1. Romania Tel. and which sustained a force equivalent to the lift to be created. starting with the wing ends. and therefore its lift. and therefore the lift. and the bound vortex system. Email: cberbente@yahoo. Instead. This high speed calls for extremely high accelerations and produces very large viscous forces. the circulation round the wing. collocation. Romania June 1 .Carafoli” University Politehnica of Bucharest Sterian DĂNĂILĂ. when supercomputers and large computer codes are available. The primary results that may be obtained from the model are values of the induced drag coefficient and the lift-curve slope of a finite wing. Fax: (+40) 21 3181007. INTRODUCTION Even today. increasing progressively as the stagnation point moves back. at the instant of starting. When a wing is accelerated from rest the circulation round it. Department of Aerospace Sciences “E. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. Fascicola 3. and then accelerates from rest in a . is not produced instantaneously.. at the sharp trailing edge the fluid is required to change direction suddenly while still moving at high speed. However a collocation type method is not the most adequate one as it has to face chord or/and incidence jumps in several wing sections. These jumps will cause instabilities altering the calculation results. In the following one presents a method able to avoid such jump instabilities by using a weighting integration over the wing span. Assoc. The differential–integral equation for the vortex intensity can be reduced to an infinite system of linear equations for the coefficients of the Fourier series of circulation. Prof. Prof.com ABSTRACT The Prandtl lifting line model is able to consider the main effects of the aerodynamic interaction between a parallel uniform flow and a wing of finite span with a sufficiently large aspect ratio. 2007 A SEMIANALITICAL METHOD FOR THE WING AERODYNAMICS Corneliu BERBENTE. aerodynamics. KEYWORDS Lifting line.* Department of Aerospace Sciences “E.2. comes to rest at the trailing edge. When the stagnation point reaches the trailing edge the fluid is no longer required to flow round the trailing edge. The vortex system can be divided into three main parts: the starting vortex. LIFTING LINE THEORY 2.

Figure1.. on the undersurface the fluid will either tend to flow inwards to a lesser extent. a representation directly related to the viscous action within the boundary layer. the circulation round the wing. the difference in spanwise velocity will cause the fluid to roll up into a number of small streamwise vortices. force systems. Figure 3. is a hypothetical arrangement of vortices which replace . effects. The hypothetical bound vortex must do the same. has become stabilized at the value necessary to place the stagnation point at the trailing edge The vortex which has been left behind is equal in strength and opposite in-sense to the circulation). due to the real aerofoil. or may tend to flow outwards. round the wing and is called the starting vortex or initial eddy (Figure. Trailing vortices Both the starting vortex and the trailing system of vortices are physical entities which can be explored and discerned and seen if conditions are right. The vortex is left behind at the point reached by the wing when the stagnation point reached the trailing edge. and may be greater than that of the surrounding environment. all the properties. For complete equivalence. distributed along the whole span. The strength of each of these two vortices will equal the strength of the vortex replacing the wing itself. disturbances. Let’s consider an incompressible flow around a wing of finite span as sketched in Figure 4. Its reaction. The varying span wise lengths accommodate the grading of the lift towards the wing-tips. These small vortices roll up into two large vortices just inboard of the wing-tips (Figure 2). at least a little distance away. 1) the real physical wing in every way except that of thickness. etc. the fluid will tend to flow inwards towards the root from the tips.2 Prandtl-Glauert formulation This model is an important example of using vortices for the representation of solid-fluid interaction by vortex singularities. Where these two streams combine at the trailing edge. A real aerofoil produces a trailing vortex system. The bound vortex system must produce the same grading of lift along the span. Bound vortex system A consequence of the tendency to equalize the pressures acting on the top and bottom surfaces of an aerofoil is for the lift force per unit span to fall off towards the tips. over the upper surface. The replacement bound vortex system must create the same disturbances. on the other hand. It is largely concerned with developing the equivalent bound vortex system which simulates accurately. and this mathematical model must be sufficiently flexible to allow for the effects of the changed parameters. Thus. Similarly. The bound vortex system. This is the essence of finite wing theory. the ends turned back produce the trailing system and the two physical attributes of a real wing are thus simulated (Figure 3). while the pressure on the lower surface is greater than that on the upper surface. The initial eddy The pressure on the upper surface of a lifting wing is lower than that of the surrounding environment. the bound vortex system should consist of a large number of spanwise vortex elements of differing spanwise lengths all turned backwards at each end to form a pair of the vortex elements in the trailing system. being replaced by fluid which was originally outboard of the tips. 2. Figure 2.different direction.

S ( π ) = ∑ ( −1) p p 2 Ap 1 ∞ has a slower convergence in terms of p2. Thus. This approach has disadvantages at discontinuity points (in chord and/or angle of attack). but small modification in the wing planform appears. K . Defining θ by: b y = − cos θ . ∞ (1 ) Here α denotes the angle of attack. p =1 ∞ (3) Now the problem consists in determination of the unknown coefficients Ap. 3. 0 π H np = H pn µ sin nθ sin pθ . 1 ∞ (6) δnp being the Kronecker symbol.. are selected on the bounded vortex line (along the y axis). The wing geometry The well known lifting line theory of Prandtl leads to the following differential–integral equation: ⎛ 1 Γ( y ) = kπcV∞ ⎜ α − ⎜ π 4 V∞ ⎝ dΓ dy ′ d y′ y − y −b / 2 b/2 c0 sin θ = β0 + 2β2 cos 2θ + 2β 4 cos 4θ c (7 ) . for example for elliptical and polygonal wings. having Ap. the term: S ( θ) = ∑ 1 ∞ (8) I n = ∫ µα sin nθ . at these points a chord discontinuity is present.…. In case of symmetrical wing and incidence α(θ) one yields: . one gets a stable rapid convergent system of equations for An : ⎞ ⎛π ⎜ + nH nn ⎟ An = I n − ∑ p (1 − δ np ) H np Ap . ∫ that smoothes the jump... . c(y) the wing chord. PROPOSED FORMULATION Present improvement has several advantages: a) it conserves the wing real chord variation.. n. c) is easier to be adapted for any kind of calculations (including numerical calculation). p = 1. p = 1. one gets: ∑ Ap sin pθ = µ⎜ ⎜α − ∑ 1 ∞ ⎛ ⎝ ∞ 1 pAp sin pθ ⎞ ⎟ sin θ ⎟ ⎠ (4 ) . as unknowns results.. By multiplying equation (4) with the weight functions sin nθ n = 1. In order to avoid the chord discontinuity at the wing ends.3. dθ =∫ sin θ 0 π (9) pAp sin pθ sin θ . p = 1. The equation is then imposed at each of these points and a set of linear algebraic equations. 2 (2) Glauert assumed Γ as a Fourier series: Γ = 2bV∞ ∑ Ap sin nθ . (5) at wing extremities S ( 0) = ∑ p 2 A p . both for the analytical and numerical calculations.2. a first improvement was done by Carafoli. There are many weight functions fitting the problem necessity. p = 1. This corresponds to a Galerkin type selection what was proved to be very adequate in many other cases..3. Under the form (4) the determination of the coefficients Ap..2. b) gives suggestions to accelerate the convergence of the infinite system of equations for the coefficients A p.. meets certain difficulties. that vanish at the wing ends. due to instabilities at end stations θ = 0 and θ = π . leading to simpler formulas. Introducing (3) in equation (1).. then integrating over the wing span. Will choose however functions of the same kind as appearing in the series (3).. µ = kc 2b A certain number of points.2. The integrals In..3.. In addition.2. then integrating along the wing span.. called collocation points. [1].2.2. p = 1.. π ] . A collocation method is equivalent to multiplying the equation (4) by a Dirac weight function centered on the collocation points.can be directly performed in many practical cases. Hnp.2. θ ∈ [0. ⎠ ⎝2 1 n = 1. by considering an approximate representation of the chord variation: Figure 4. V∞ the upstream undisturbed velocity and Γ( y ) the circulation [2].

λ = b2 . p (see relations (15)). 2. provided with a slat. 2 p − 3+ 4 µ 0 ( 2 m − 1) − ( 2 m − 1) 2 − 4 ( p − 1) 2 A consequence of relations (10). p≠m ∞ 2m−1. one gets: A1 = µ 0α0 = (24) 30π( 2 − q ) + 16µ 0 (64 − 72q + 15q 2 ) 2 15π + 48πµ 0 (7 − 5q) + 16µ 0 (64 − 72 q + 15q 2 ) 2 (16) In this way one can calculate wings of various shapes.2 p −1 = H 2 p −1.. A3 = 1 (I 3 − H 31 A1 ) 2 L3 L1L3 − 3H13 lim ( m − 1)(1 − cos mπ cos pπ = 0 . The lift coefficient results: C z = πλA1 . Cx . the integrals H2m–1.2 p −1 = H 2 m −1.…. This circumstance could appear in a wind tunnel with rectangular section and for a sail (Figure 5) fixed on a mast at a distance ε /2 from the deck of a vessel (practically such rigid metal sails are used to maneuver big vessels near the shore). respectively. respectively. For these reasons one may truncate the infinite sum in (10). 3. 2 p−1 A2 p−1 . rigid connected to the mast and. In Figure 5 a hybrid combination between a soft sail made of linen and a solid sail.2p – 1 are decreasing with the indices m. m = 1.. Thus we can restrict ourselves to give only the coefficients A1 and A3: A1 = I1L3 − 3I 3 H13 . 5.3.K ( m − 1) 2 − ( p − 1) 2 (23) For the first coefficient A1. A3. using three equations corresponding to m =1. α(θ) = α(π − θ ) . S (17) where S and λ are the reference wing surface and the wing aspect ratio respectively. ASYMMETRIC WING NEAR WALL In the following one considers a wing not necessarily symmetric of span b/2 at a distance ε /2 with respect to an infinite vertical plan. we neglect in (14) terms A2 m + 3 with respect to A2 m −1 (four indices distance). starting with p = m + 2. 2 m −1 where: q = 1− ce . (12) is the vanishing of the coefficients An for even values of n: A2 m = 0 . A3.. whereas the geometrical angle αg is reported to the x axis. (10) (11) (12) H np = 0 . For a better accuracy. the aerodynamic forces can be evaluated. p = 2... n n ≈ z ⎜1 + 3 3 πλ ∑ A2 πλ ⎜ A2 ⎟ n =1 1 ∞ A2 C2 ⎛ ⎝ A2 ⎞ 1 ⎠ (18) 4. The coefficient of the induced drag . for n + p = odd. 2b c0 (21) where: Ln = π 2 (22) nH nn (15) Taking into account that: p→m Under assumption (subsequently confirmed) of a rapid convergence of series (3) of the circulation. In = 0 . where the presence of engines and fuselage can also be considered. TRAPEZOIDAL WING For trapezoidal wings (Figure 4) the integrals above introduced can be evaluated in closed form: . The positioning angles of slat and the soft sail are ϖ v and β.1 2 2m − 1 + cos mπ . The angle of attack α0 is measured with respect to the zero lift axis.2. is: C xi = 2 Cz ⎟. On the other hand. = = −q µ 0α0 µ0 2m − 1 2m( m − 1) (19) for n = 2p = even. ⎛ ( m − 1)(1 − cos m π cos p π − qµ 0 ⎜ + ⎜ ( m − 1) 2 − ( p − 1) 2 ⎝ m (1 + cos m π cos p π ⎞ ⎟ + ⎟ m 2 − ( p − 1) 2 ⎠ (20) (13) Then one needs only the system (8) for odd indices in the form [3]: L2m−1 A2m−1 = I 2m−1 − ∑(2 p − 1)H p=1. Knowing the coefficients A1. one can start with three coefficients: A1. µ 0 = kc0 . (14) H 2 m −1.2.3.K H 2 m −1 . (11).µ(θ) = µ(π − θ) . m = 1. I 2 m −1 H 2 m −1. A5.

L3 = + 3µ0H33r . π]) . especially for larger values of the aspect ratio λ. one can assume an almost constant slope of the curve Cz(α) for a fixed configuration. α0. µ. The angle α0 is measured with respect to the null lift axis of the hybrid wing system and depends on the curvature of the profile “profiled mast–soft sail–slat”. ε /2 ) out of a complete wing with chord c0 at y = 0. the resulting motion takes place around a wing having a double span b. ⎥ b b ⎢ ⎣ 2⎦ A2 = 0 . For the symmetrical wing the chord parameter variation. At y = ± ε / 2 one have a new jump in the chord distribution. As a conclusion. both with big slat (VP1a) and small slat (VP2a). As a consequence. The rectangular hybrid wing (VP) and the trapezoidal hybrid wing (VT) The system wing–solid infinite wall is equivalent from aerodynamic point of view with a „wing in a mirror”. one may see a significant displacement (26) Thus. defined by: sin β ε = ε π⎤ . The other integrals are expressed as functions of I1r and I2r. are calculated by extracting the portion (–ε /2.the angle of the relative speed V∞. The simplest case corresponds to elliptical wing. Although this dependence is complicate. b) the trapezoidal hybrid wing. β ε = arcsin ε ∈ ⎡ 0. q = 1 − e . Once more the proposed method is recommended. Therefore the assumption of a almost unmodified direction of the null lift axis direction is confirmed. suggest a linear variation of CZ with α0 . with big slat (VT1a) and small slat (VT2a). The rigid mast was profiled as a wing. the lifting line theory is applicable for angle of attack α0 corresponding to interval (CZmin. (30) Figure 5. By denoting with βε the angle. We have calculated the required integrals for a trapezoidal wing. In exchange. one obtains for the integrals I2m-1. CONCLUSIONS The good agreement between theory and experimental data for the slope dCZ/dα0. µ 0 = 0 . defined by relation (4): kc c kc µ= = µ 0 (1 − r cos θ ) . CZmax). at constnt incidence. Then one gets as particular cases the rectangular and the triangular wings. Comparisons of the analytical calculations with experiments are given in Tables 1 and 2. µ 0 α0 (27) . 2b 2b c0 We have taken in consideration four cases: a) the square hybrid wing. defined by relation (14): I1 ≡ I 1r = 2(1 − sin β ε ) − r cos 2 β ε . the four cases presented in Table 1.I3 2 ≡ I 3r = (1 + sin 3β ε ) − r 1 + 2 sin 2 β ε cos 2 β ε µ 0α0 3 ( ) the index „r” standing for „reported”. as follows: L1 = π π + µ 0 I 1r . 6. A3 I ≡ A3r = 3r (1 − µ 0 A1r ) µ 0 α0 L3 (25) Integrals along the wing span (θ ∈ [0. 2 2 (28) ⎛ 6 sin 2 5β ε ⎞ − sin β ε ⎟ H 33r = I 3r + 2⎜ ⎜5 − ⎟− 5 ⎝ ⎠ ⎛ 4 sin 2 3β ε ⎞ − r⎜ − sin 2 β ε ⎟ ⎜3− ⎟ 3 ⎝ ⎠ (29) Now one can calculate the first three coefficients of circulation: 2 A1 I L − 3µ 0 I 3 r ≡ A1r = 1r 3 2 2 µ 0 α0 L1 L3 − 3µ 0 I 3r .

10 theory (indus) 0.0 17. 3.277 0.a. Carafoli E.650 0.4] (min error) 0..0769 exper. 60%) is induced by the free vortex sheet. Constantinescu V.7078 0. The proposed method for solving the Prandtl equation.50 αimax (deg) 10.7208 0. 2.7 -12. VP2a VT1. Table 1. Series D. The two times increases of the angle α0max in comparison to usual wings can be explained through the slat effect as well by flexible wing adaptation to flow (Table 2). As well.1164 0.423 0.0 8.0682 0.25 2.Acad. (2005) Experimental 4. Asupra unei metode pentru calcul aerodinamic al aripilor de anvergurǎ finitǎ în regim incompresibil.Cx. – a convenient arrangement of the system of equations giving the coefficients A1 . VP2.4%) (62. Bul.834 6.. Ed. the angle α0 at Czmax keeps a value around 30 degree (Table 2).500 0.P.a VT2.4418 λ 3. Carafoli.25 2.4030 0.of null lift axis with wing planform.500 (57. U.682 0.8000 µ0 0.0%) Table 2. of the „wing in mirror”).920 5.0769 (9. the rest belonging to friction effects.0751 0. C.41 10. R..4298 0.0627 0. Researches for Determining the Propulsion Parameters of the Hybrid-sail Ships..05 2.5027 0.0789 0. 5.0 30. pp.32 .750 0. pp.689 3.a VT1.5 -23. C.7568 0. Dinamica fluidelor incompresibile.05 7. Inst.491 A1r 0. E. eventually A5 (A2 = 0 .372 0.4%) (55.. (1945). more than a half (approx. A3 .S. Comparisons analytical model with experimental data for the slope of lifting coefficient Model VP1. As regards the drag coefficient. Politehnic Bucureşti. – permits an exact calculations of the integrals leading to simple analytical expressions.77 7.a VT2a r 0.5 31. 0. but not with the slat chord. (1973).0185 -1 dCZ / dα 0 (deg ) theory 0. Berbente C. at Czmax. Theoretical and Experimental Study Regarding the Aerodynamics of a Ship Sail System.7 29.a.0 α Czmax 7. Série de Mécanique Appliquée.B.310 experiment[3.R. has the following advantages: – it takes into account without problems the sudden chord variations or jumps (four jumps in case.2%) (56. Analele Acad. Bucureşti.0276 0.0 18.1235 0.0 -12. Theorie des ailes monoplanes d’envergure finie.4670 0.17 . XXXV.0%) REFERENCES 1. 50. 37-49. Revue Roumaine des Sciences Techniques. N.Române. Maraloi.0 Czmax 2. Scientific Bulletin.0 δα (deg) 30.4] 0.0614 0.a α (deg) Cz = 0 -23.[3. (2006). Berbente. 68.82%) (0..4941 A3r 0. Bucureşti. Maraloi C. A4 = 0) with a good accuracy.94%) (8.09%) (4. 0. Comparisons analytical model with experimental data for Cx for Czmax Cx for Czmax Model VP1. (1981).

. Further the Euler equations for axi-symmetric flows are simplified using a number of assumptions based on the slenderness of the vortex core and on the conical flow structure inside the vortex. 2007 A PHYSICAL AND THEORETICAL ANALYSIS OF THE VORTEX BREAKDOWN ON DELTA WINGS Marius STOIA-DJESKA. resembling the jet-like stable region and the wake-like after breakdown region of the flow. This is in fact a well-ordered vortical flow structure. The resulting equations can be solved in closed form and the solutions obtained show two different features. E-mail: marius. Depending on the Mach number and geometrical parameters there is a critical angle of attack at which the internal structure of the vortex changes substantially and the vortex breakdown occurs. vortex breakdown also marks the end of the favorable effects induced by the leading edge vortex. Department of Aerospace Sciences “Elie Carafoli” “POLITEHNICA” University from Bucharest Corneliu BERBENTE. Assoc. KEYWORDS Vortex breakdown. Vortex breakdown means a sudden increase of the cross section of the vortex core and the loss of the regularity of the flow. Prof. The leading edge vortex occurs when the flow encounters a sufficiently highly swept sharp leading edge and can be explained within the framework of inviscid flow theory by the Kutta condition.* Department of Aerospace Sciences “Elie Carafoli” “POLITEHNICA” University from Bucharest Sterian DANAILA. These features are discussed in the paper. Fascicola 3. The resulting equations subjected to rational boundary conditions can be solved in closed form and the solutions obtained show two different features. some experimental results are presented in order to explain the vortex break down phenomenon. vortex flow on delta wings NOMENCLATURE vr [m/s] radial velocity vθ u α M [m/s] [m/s] [deg] [-] tangential velocity axial velocity angle of attack the Mach number Subscripts and Superscripts e external conditions 1.ro wings. The phenomenon of vortex breakdown has been observed in wind tunnel experiments and real flights and can be investigated using computational simulations as well as theoretical approaches. some experimental results are presented in order to explain the vortex break down phenomenon. Romania June 1 .: (+40) 21 4023967. Prof. The low pressure induced on the upper side of highly swept wings by the leading edge vortices results in an important increase of the lift capabilities of small aspect ratio . 5]. This non-linear effect was and is intensively exploited in the design of modern civil and combat aircrafts [1.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). 4. Prof.stoia@rosa.2. Department of Aerospace Sciences “Elie Carafoli” “POLITEHNICA” University from Bucharest *Corresponding author: Polizu 1-6. Further the Euler equations for axially symmetric flows are simplified using a number of assumptions based on the slenderness of the vortex core and on the conical flow structure inside the vortex. First. Increasing the angle of attack of the wing leads to an increase of the strength of the leading edge vortex. steady and stable. Romania Tel. INTRODUCTION In aircraft aerodynamics and not only vortex flows play an important role. Bucharest. First. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. 011061. This paper concerns with the analysis of the stability of a narrow vortex core. resembling the jet-like stable region and the wake- ABSTRACT The paper concerns with the vortex breakdown phenomenon. Further. Fax: (+40) 21 3181007.

2. a) α = 10deg. 5]: • The presence of the stable primary vortex strongly affects the pressure and velocity distributions on the wing upper side. • This suction region gives the so called “vortex lift” effect that represents an essential feature of highly swept delta wings. χ LE = 650 Wind tunnel studies and numerical simulations have proven that [2.like after breakdown region of the flow. The oil flow visualization of vortex breakdown at M = 0. d) Starboard vortex breakdown at α = 18. These features are discussed in the paper.3 deg. . The vortex has a viscous core but the flow outside the core is essentially inviscid. FUNDAMENTALS OF DELTA WING AERODYNAMICS AND VORTEX BREAKDOWN The subsonic flow over a sharp edged delta wing at medium and large angles of attack is characterized by leading-edge separation. Due to this embedded vorticity the free shear layer rolls up in a spiral fashion and the result is a steady primary vortex. 4. As the angle of attack increases over a limit value a large scale vortex breakdown occurs . Figure 1. c) Vortex breakdown is imminent at α = 18. 3. the predominant and beneficial effect being a low-pressure region under the vortex core. b) α = 18 deg. This secondary vortex is due to the flow separation due to strong adverse pressure gradients in the spanwise direction. The flow separates when encountering the sharp leading edge and forms a free shear layer which contains vorticity. The dominant aspect of the flow is the vortical structure that forms on the upper side of wings with leading edge sweep angles greater than 45 deg.85. The vorticity from the entire length of the leading edge is convected through the free shear layer to the vortex core which increases in strength and in cross-section in the downstream direction.5 deg. There is also a second vortex sheet lying above the primary vortex.

We consider therefore that a better understanding of the vortex breakdown mechanism represents the first step towards new solutions for the flow control. Introducing these partial derivatives in (2) one obtain a new system of partial differential equations that contains product terms like R ∂u ∂u ∂p or R . At lower angles of attack the primary vortex burst point is downstream the trailing edge and does not affect the vortex lift. 3. Detailed measurements show that the axial component of the velocity is increasing up to the breakdown point location (axial velocities more than three times the freestream velocity were measured) where it decreases abruptly. It means a sudden increase of the cross section of the vortex core and the loss of the regularity of the flow. In what follows we assume that the vortex core is isolated and its cross-sectional dimension is small relative to typical length scale along the axis of vortex core. For non-conical flows but for slender and narrow vortex cores terms like those above can be neglected.1. The vortex burst or breakdown mechanism is one of the unresolved problems in fluid mechanics. . Then the pressure partial derivatives become: ∂p 1 ∂p dp 1 dR ∂p ∂p = . Simplified equations of the vortical flow It is now clear that the vortex breakdown mechanism is inviscid and thus the incompressible Euler equations can be used as the starting point for searching for rational but simplified flow models. Further. Upstream the bursting point the core has a jet-like structure while downstream the core has a wake-like structure that expands in the radial direction. This assumption we made means that the vortex core is slender and slowly varying in axial direction. narrow region with rotational axially-symmetric flow the equations reduce to: ∂vr ∂u 1 + + vr = 0 ∂x ∂r r ∂u ∂u 1 ∂p + vr + = 0 u ρ ∂x ∂x ∂r ∂vr ∂vr 1 1 ∂p + vr − u θ2 + = u ρ ∂r ∂x ∂r r ∂ uθ ∂ uθ 1 u + vr + u r uθ = 0 r ∂x ∂r ( 2) The unknowns of these equations are the velocity components and the pressure.above the wing and the flow becomes fully unsteady. bursting involves a sudden decrease in the magnitude of the axial and circumferential velocities of the core. TU Delft. The interest is now to control the flow so that the bursting point to remain downstream the trailing edge for medium and high alpha. Ru ∂x ∂x ∂x the quasi-conical vortex core the velocity and the pressure do not depend on the axial direction. These can be expressed as functions of x and the dimensionless radius r = r R ( x ) where R ( x ) denotes the radius of the cross-section of the core for x=const [3]. there are at least two completely different states of the vortex core that must be identified.85. MATHEMATICAL MODEL AND CLOSED FORM SOLUTIONS 3. Further. for aeronautical applications this approach is inefficient. χ LE = 650 obtained at High Speed laboratory. Unfortunately. Figure 1 shows some results of the oil flow visualization study of vortex breakdown on delta wings at M = 0. in other words by increasing the jet like axial velocity. For conical flows inside . It was discovered that the vorticity vector is first oriented in the axial direction and after the vortex breakdown reorients itself along a transverse direction [4. The white region represents approximately the low-pressure area and also delineates the position of the primary vortex. 2]. It is nowadays known that the breakdown can be postponed by axial air blowing or. If we do so then the system (2) reduces to a system of ordinary differential equations with the only independent variable r . The solution of the equations gives a qualitative insight in what happens inside the vortex core. in the slender. For incompressible inviscid flow the Euler equations in cylindrical coordinates are: ∂u ∂vr 1 1 ∂vθ + + vr + =0 ∂x ∂r r r ∂θ ∂u ∂u 1 ∂u 1 ∂p + vr + + =0 u ∂x ∂r r ∂θ ρ ∂x (1) ∂vr ∂vr 1 ∂vr 1 2 1 ∂p + vr + − uθ + =0 u ∂x ∂r r ∂θ r ρ ∂r ∂u ∂u 1 ∂vθ 1 1 ∂p u θ + vr θ + + u r uθ + =0 r ∂θ r ∂x ∂r ρ r ∂θ Under the assumptions we made. = −r + ∂r ∂x R ∂r dx R dx ∂r (3) and the same rule applies for the velocity components..

(1992) High Angle of Atack aerodynamics. Closed form solution The boundary conditions we assumed lead to the following closed solution u = U e (1 − V ⋅ H ( r ) ) vr = −V ⋅ U e G ( r ) ⎡ V2 vθ = SU e ⎢1 − H (r ) + H 2 (r ) − 2 ⎢ ⎣ S p − pe ⎡⎛ V ⎞ V = V ⎢⎜1 + ⎟ H ( r ) − I ( r 2 2⎠ 2 ⎣⎝ ρU e velocity and vorticity vectors are pointing in opposite directions.B. (1978) The Aerodynamic Design of Aircraft. REFERENCES 1. CONCLUSIONS We have examined the physical behavior of the vortex flow on a delta wing with sharp and highly swept leading edge. Vol. For W = −1 the .85. Kuchemann. 209-227 4. Hall.e. (1961) A Theory for the Core of a Leading Edge Vortex. (1990) Numerical Simulation of the Effects of variation of Angle of Attack and Sweep Angle on Vortex breakdown over delta Wings. L. Pergamon Press 2. Oil flow visualization of vortex breakdown on delta wings at M = 0. 3. Sciff. pp. Technical University Delft. When the axial component of the velocity becomes positive the spiraling streamline starts to run downstream but with a sudden increase of the spiral radius. J.G. AIAA -90-3000-CP 3. At the exterior boundary of the vortex core the axial and circumferential components of the velocity and the pressure match continuously the outer flow values. The components of the vorticity vector can now be derived analytically too and is left to the reader. p = Pe ( x ) for r = 1. The jet-like solution develops in the upstream of the bursting point while the wake-like solution develops downstream this point.2.e.. 11.E. vθ = Ve ( x ) . J. Ekaterinas. i. This is the wake-like part of the solution and this can be observed on the rear part of the wing.3. The solution has two branches. Further we derived an analytical solution for an isolated slender vortex core. M. J. NATO RTO AVT-E-10/RTG 5.. Springer Verlag. D. Lamar.A. A streamline coming from upstream enters the vortex core and continues downstream spiraling around the core axis with a slight reduction of the spiral radius. vr = 0 for r = 0. The behavior of the solution (4) depends on the values of the parameter W. We consider that the vortex breakdown mechanism is in fact responsible for the switch in the solution state. JFM. The streamline starts at infinity downstream close to the axis and runs upstream with an increasing spiral radius. 4. New York ( ( 1) (4) where we denoted the functions: G (r ) = (1 + r ( 2 R′ 2 ) 1 2 −1 . the mechanism responsible for the switch in the flow state from one to the other is still unclear. u = U e ( x ) . ACKNOWLEDGMENTS The present work has been supported from the National University Research Council Grant (CNCSIS) 33/2004. (2002) Understanding and Modeling Vortical Flows to Improve the Technology Readiness Level for Military Aircraft – terms of reference. For W = 1 the velocity and vorticity vectors point in the same direction. Rom. Boundary conditions At the axis of the core the radial velocity vanishes.3. This is the jet-like part of the solution and it appears in the forward part of the wing. However. χ LE = 650 have been obtained from High Speed laboratory. ⎛ ⎛ 1 + R ′2 ⎜r⎜ ⎜ ⎜ ⎝ ⎝ r R′ ⎛ H ( r ) = ln ⎜ 1 + r 2 R′2 ⎜ ⎝ I (r ) = H (r ) − 2 ) 1 2 ⎞ − 1⎟ ⎟ ⎠ ( ) ) 1 2 ⎞⎞ − 1⎟ ⎟ ⎟⎟ ⎠⎠ −1 ( 1 + r 2 R ′2 ) 1 2 −1 r 2 R ′2 + ( 1 + R ′2 1 2 R ′2 and V =W R'2 ⎛ 2⎜ 1 + ⎜ ⎝ 1 ⎡ ⎤ 1 ⎞2 ⎥ ⎢⎛ '2 2 1+ R −1 2 ⎟ ⎢⎜ ⎥ S ⎟ − 1⎥ (5) ⎜1 + 4 '2 ⎢ ⎞ ⎜ R ⎟ ⎥ ⎜ ⎟ − 1⎟ ⎢ ⎝ ⎠ ⎟⎢ ⎥ ⎠⎣ ⎦ ( ) ( 1 R'2 2 ) The swirl ratio of the flow outside the vortex core is denoted with S = Ve U e and the parameter W = ±1 . i.

cavitation tip clearance. The runner blades were manufactured with anti cavitation lips. No. d. Geometrical model The 3D blade model without and with anti cavitation lip is done in the figure 1 and 2.* Faculty of Engineering “Eftimie Murgu” University of Resita Dorian NEDELCU.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). Experimental investigations were performed on the model in the laboratory and on the prototype in power station. Resita.2. . Faculty of Engineering “Eftimie Murgu” University of Resita *Corresponding author: P-ta Traian Vuia.nedelcu@uem.ro. 320085. Romania Tel. beginning from the leading edge. The 3D model of the blade without anti cavitation lip Figure 2. NUMERICAL SIMULATION AND EXPERIMENTAL RESULTS Constantin Viorel CAMPIAN.ro ABSTRACT The paper presents a flow numerical simulation and experimental results regarding cavitation tip clearance for Kaplan turbines. NUMERICAL SIMULATION RESULTS 2. Tip clearance cavitation appears inside the gap between rotating blade and the runner chamber. Romania June 1 . in two geometrical solutions. Fascicola 3.: (+40) 256 219134. Email: v.campian@uem. Supplementary. Assoc. The 3D model of the blade with anti cavitation lip 2. where the second solution correspond to a shorter length of the anti cavitation lip comparative with first solution. Experimental results are obtained on the model and prototype. INTRODUCTION Improvements in hydrodinamics design of hydraulic turbine must generate a better control of cavitation phenomenon. In this paper cavitation tip clearance is analyzed. When the boundary layer separation occurs. The analysed domain correspond to the gap between runner blades and runner chamber. 2007 CAVITATION TIP CLEARANCE. which is the source for the vortex cavitation. Figure 1. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. KEYWORDS Kaplan turbine. Numerical simulation of the clearance water flow was made by Navier-Stokes method. Fax: (+40) 256 219134. the tip clearance cavitation appears. vortex flow 1. Prof. 1-4. The flow through the clearance is the results of the intrados/extrados pressure difference. a vortex attached to the blade tip is generated in the channels between the blades. Prof.1. For anti cavitation lip were analyzed two geometrical solutions.

2. Numerical simulation The numerical simulation were performed with Navier-Stokes method [1] for blade without and with anti cavitation lip in two solutions. 3 to 6 for solution 1 and 7 for solution 2. Figure 6 Pressure distribution on the blade with anti cavitation lip solution 1 – cavitation bubles near blade surface Figure 3 Pressure distribution on the blade without anti cavitation lip – top view Figure 7 Pressure distribution on the blade with anti cavitation lip solution 2 – top view Figure 4 Pressure distribution on the blade without anti cavitation lip – front view Figure 8 Comparative pressure distribution on the blades with anti cavitation lips The area close to the anti cavitation lip is the most important area for this study.2. Figure no. because the cavitation pitting area will be smaller. 8 presents comparative results between the two solutions. Figure 8 shows that solution 2 for anti cavitation lip (with shorter length) is better than solution 1. Figure 5 Pressure distribution on the blade with anti cavitation lip solution 1 – top view . The numerical results are graphically presented in figures no. The pressure distributions are calculated at the same head for suction side of the blade where the anti cavitation lip is placed.

Experimental results on the prototype The experimental observations on the prototype runner blade were made after 8000 hours of operation. The efficiency and the cavitation picture on the model blade are the same without and with anti cavitation lips (solution 1 and 2). the tests were performed.1. 3.2. an area with depth cavitation pitting near the anticavitation lip appears. figure 10 and 11. On the blades equiped with this solution of anticavitation lips. at the same operating conditions. blades with long (solution 1) and short (solution 2) anti cavitation lips. In the figures 12 and 13 are presented pictures with cavitation pitting areas and the long anticavition lip (solution 1). Figure 11 Blade with anti cavitation lip solution 2 . figure 9. On the blade near the anti cavitation lip was observed a slightly cavitation erosion area. Experimental results on the model Experimental results were performed on the Iron Gates I turbine model [2].3. Figure 9 Blade without anti cavitation lip Figure 12 Cavitation pitting on the anti cavitation lip – solutia 1 Figure 10 Blade with anti cavitation lip solution 1 Figure 13 Zoom on the cavitation pitting on the anti cavitation lip – solutia 1 In the figure 14 is presented a picture with cavitation pitting areas and the short anticavition lip (solution 2). for 3 blade configurations: blades without anti cavitation lips. EXPERIMENTAL RESULTS 3. The model has been installed on the test rig. which is not visible on the figures.

Result Presentation in Zurich. solution 2 of the anticavitation lip induce on the lip and blade a significance cavitation erosion. 2. On the prototype. This cavitation behaviour was not possible to be observed on the model. VA TECH – CFD Cavitation Research. The tip clearance cavitation analysed in the study can not be numerical reproduced with accuracy. April. which generate the cavitation erosion on the blade. CONCLUSIONS There are no significant cavitation differences between the blades with long. 2003 . June. VA TECH – PdF1 Report of the Task Force Cavitation. comparative with solution 1. 2002. at the model tests. REFERENCES 1.Figure 14 The cavitation pitting on blade and the anti cavitation lip – solutia 2 4. The very small gap (5 mm) between the runner blade and runner chamber compared with the runner diameter (9500 mm) and the size blade compared with the small dimensions of the anticavitation lip limit the approximation of the phenomen by numerical simulation. short and without anti cavitation lip. The modified geometry of the solution 2 (short length) determine the peripheral vortex attachment.

Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). Figure 1. Prof. rigorously speaking. Department of Hydraulic Machinery “Politehnica” University of Timisoara Alexandru BAYA. Prof. a the flow and the boundary conditions imposed and simplified simulation technique must be employed in the end the results of the flow simulation. Prof. to obtain useful results for pump analysis. Prof. Department of Hydraulic Machinery “Politehnica” University of Timisoara *Corresponding author: Bv Mihai Viteazu 24. INTRODUCTION The progress in the field of Computational Fluid Dynamics (CFD) has made this technology an important tool in analysis and design of hydraulic turbomachinery. Assist. The turbomachinery flow is essentially unsteady due to the rotor-stator interaction. Fascicola 3. Senior Researcher Center of Advanced Research in Engineering Sciences Romanian Academy . Romania June 1 . First the investigated centrifugal pump requires large computer memory and CPU pump is described.mec. the geometrical periodicity of the rotor blade channels cannot be used since there are differences in the flow from one inter-blade channel to another. Email: astuparu@mh. using currently available computing resources.ro validated assumptions. then the equations that governs time even for our days computers. Fax: (+40) 256 403700. EQUATIONS AND BOUNDARY CONDITIONS The investigated storage pump has two stages with the impellers situated in opposition and a suctionelbow of complex geometry. numerical investigation. However.: (+40) 256 403692. with carefully chosen and experimentally Figure 1. As a result.0.* Department of Hydraulic Machinery “Politehnica” University of Timisoara Daniel BALINT. computing the real flow of a storage pump by using commercial code (unsteady and turbulent) through the whole storage FLUENT 6.Timisoara Branch Liviu ANTON.2. Romania Tel. the 3D flow in the impeller and the suction-elbow However. Timisoara. On the other hand. Cross section through the storage pump . KEYWORDS storage pump. 2007 NUMERICAL ANALYSIS OF PUMP HYDRODYNAMICS AT CONSTANT SPEED Adrian STUPARU. Each impeller has five blades. Assist. Figure 2. cp = g p − p IN [-] ρgH gravity pressure coefficient [m/s2] [m2/s2] k turbulent kinetic energy 1. Department of Hydraulic Machinery “Politehnica” University of Timisoara Sebastian MUNTEAN. turbulent flow NOMENCLATURE 2. so that very good and engineering useful This paper presents the numerical investigation of results are obtained [5]. 2007 3rd Workshop on Vortex Dominated Flows Bucharest. COMPUTATIONAL DOMAIN. 300223.upt. one can develop a ABSTRACT methodology for computing the turbomachinery flow.

As a result. the viscosity coefficient is written as a sum of molecular viscosity µ and turbulent viscosity µT. but different in angular extension) and extends downstream up to cylindrical patch. W =V −ω×r (2) with r the position vector. Isometric view of the storage pump impeller The computational domain includes only the impeller of the centrifugal pump. By introducing the relative velocity Figure 3. and the last term in the righthand-side of (1b) becomes ∇ ⋅ [(µ + µT )∇V ] . the left hand side of (1b) becomes ∂ (ρW ) + ∇ ⋅ (ρWW ) + 2 ρω × W + ∂t ∂ω + ρω × (ω × r ) + ρ ×r ∂t (3) An important assumption used in the present computation is that the relative flow is steady. see Figure 4. when only the impeller domain is considered. [3]. The computational domain is bounded upstream by an annular section (wrapped on the same annular surface as the suction outlet section.0. in order to impose the boundary conditions on the outlet section. using a Reynolds-averaged Navier-Stokes (RANS) solver. ∇ ⋅V = 0 d (ρ V ) = ρg − ∇p + µ∆V dt (1a) (1b) The numerical solution of flow equations (1a) and (1b) is obtained with the expert code FLUENT 6. Isometric view of the inter-blade channel of the impeller The inter-blade channel domain is discretized with 500k cells using a structured mesh coupled with a special boundary layer discretization on the blade faces. we solve a relative flow. For the first case. Mesh generated on the 3D computational domain of the inter-blade channel For the flow analysis presented in this paper we consider a 3D turbulent flow model. For the numerical investigation only one inter-blade channel is used because of the symmetry of the geometry. This . Figure 4.Figure 2. in a rotating frame of reference with angular speed ω = ωk ( k being the unit vector of the pump axis direction). Figure 3 shows the 3D computational domain with boundary conditions corresponding to an inter-blade channel of the impeller. A steady relative 3D flow is computed.

simplifies (3) by removing the first and last terms.θ + n .42 150 3.2 power W] 73 Rotational n [rot/min] 1500 speed Flow rate Q [m3/s] 1. ⎝ ⎛ 2π ⎞ p(r . θ . and also allows the computation of impeller flow on a single inter-blade channel. z ⎟ b ⎠ ⎝ ⎛ p v2 ⎞ r r ⎜ ⎟(v ⋅ n )dS + z + ∫outlet ⎜ ρ⋅g 2⋅ g ⎟ ⎝ ⎠ H= − r r ( ) v ⋅ n dS ∫ outlet (5) ⎛ p v2 ⎞ r r ⎟ ⎜ z + + ∫inlet ⎜ ⎟(v ⋅ n )dS ρ g g ⋅ 2 ⋅ ⎠ ⎝ − r r ∫ (v ⋅ n )dS inlet (6) We investigated five operating points of the pump with the characteristics given in the table 1: Table 1: Operating points of the centrifugal pump Operating Parameter Symbol Value point Rotational n [rot/min] 1500 speed 3 Flow rate Q [m /s] 1 1 Pumping H [m] 319 head Hydraulic Pu=ρgQH[k 3128.9 8 1500 0. z ) = p⎜ ⎟.2 143.1 2 Pumping H [m] 300 head Hydraulic Pu=ρgQH[k 2941. On the inlet surface of the impeller a constant velocity field was imposed normal on the surface.61 1.1 Hydraulic Pu=ρgQH[k W] 35 power Rotational n [rot/min] 4 1500 speed 3 Flow rate Q [m /s] 0.7 168 5.62 159.θ + n . z ) = v ⎜ ⎟ ⎜ r. On the periodic surfaces of the impeller the periodicity of the velocity.8 350 3432.9 177. K ⎜ r.2 7 For the outlet section the outflow condition is imposed.2 3 Pumping H [m] 265.57 132.48 0.59 175 5.9 power W] 5 Rotational n [rot/min] 1500 speed 3 Flow rate Q [m /s] 1.17 .5 5.46 1 167. θ . The turbulent viscosity is computed using the RNG model.8 184. That means both flow and turbulent quantities remain unchanged downstream to the outlet section.45 head 2603.09 1.1 155.9 Pumping H [m] 336 And the error is calculated with the equation (7): ε= H calculat − H cata log H cata log ⋅ 100 (7) The results of the numerical simulation compared with the ones given by the pump producer are presented in table 2 and figure 5: Table 2: Pumping head values resulted from the numerical simulation and given by the pump producer HFLUENT Hcatalog ε Q [m] [m] [%] [m3/s] 0.725 8. z ⎟ b ⎠. The velocity magnitude is computed using the flow rate of the operating point: 5 vI = Q SI (4) head Hydraulic power Rotational speed Flow rate Pumping head Hydraulic power Pu=ρgQH[k W] n [rot/min] Q [m /s] H [m] Pu=ρgQH[k W] 3 3294. NUMERICAL RESULTS The pumping head is determined using the following equation: r r⎛ 2π ⎞ v (r . pressure and turbulence parameters were imposed: 3.

95 1 1.30 1.9 177.7 86 91.8 0.05 1.1 Q [m3/s] 1. Efficiency vs.42 85 92. flow rate for the investigated operating points (∫ η= (∫ − outlet ω ⋅ r ⋅ vu ⋅ ρ ⋅ v ⋅ ndS r r inlet M ⋅ω r r ω ⋅ r ⋅ vu ⋅ ρ ⋅ v ⋅ ndS M ⋅ω )− The pressure coefficient is calculated with the following relation: cp = p − p IN ρgH (9 ) (8) The results of the numerical simulation compared with the ones given by the pump producer are presented in table 3 and figure 6: Table 3: Efficiency values resulted from the numerical simulation and given by the pump producer Q HFLUENT ηcatalog ηFLUENT [m3/s] [m] [%] [%] 0.85 0.75 0.9 0.3 50 0.59 83 88.15 1. .1 Q [m3/s] 1.12 1 167.3 Figure 5.05 1.25 1. Pumping head vs.8 0.2 143. flow rate for the investigated operating points The efficiency of the impeller is determined using the equation: Figure 6.65 1.95 1 1.2 1.9 0.08 The distribution of the pressure coefficient along the pressure side and suction side of the blade for all five operating points are presented in figure 7 to 11 .300 H(FLUENT) H(catalog) 250 100 90 200 80 eta [%] H [m] 150 eta(FLUENT) eta(catalog) 70 100 60 50 0 0.57 80 93.1 155.25 1.85 0.62 87 90.15 1.2 1.8 184.

Pressure coefficient distribution along the blade for Q = 0.Figure 7.8 m3/s Figure 8.9 m3/s Figure 9.0 m3/s . Pressure coefficient distribution along the blade for Q = 1. Pressure coefficient distribution along the blade for Q = 0.

2 m3/s From figure 7 to 11 one can observe that the minimum value of the pressure coefficient is located on the leading ege and this value is decreasing while the flow rate goes up. The NPSHr (Net Positive Suction Head required) is defined as: In figure 12 the NPSHr distribution on the impeller blades is plotted for the five investigated operating points. ⎛ p V2 ⎞ p NPSH r = ⎜ i + i ⎟ − ⎜ ρ ⋅ g 2⋅ g ⎟ ρ ⋅ g ⎝ ⎠ (9) where the variables with subscript (i) correspond to the averaged one on inlet section of the impeller.1 m3/s Figure 11. It is well known. The value of the NPSHr rises with the rise of the flow rate and so the risk of cavitation is higher. Pressure coefficient distribution along the blade for Q = 1. the maximum risk of cavitation occurs where the largest NPSHr value (corresponding to minimum pressure) is obtained.Figure 10. . Pressure coefficient distribution along the blade for Q = 1. the maximum NPSHr value appears at the junction between blade suction side and the crown near to the leading edge (the red spots in Figure 12). Due to the sharpness of the leading edge.

Figure 12 NPSHr on the blades of the impeller for Q = 0.2 m3/s .8 ÷ 1.

CONCLUSIONS The paper presents a numerical study of the 3D flow in the impeller of a storage pump. Sherbrooke. El Hajem.. Pasadena. Proceedings of ASME FEDSM´01 2001 ASME Fluids Engineering Division Summer Meeting New Orleans. Proceedings of IAHR. R. A particular attention is paid to the analysis of the flow on the impeller blade.. Sallaberger. M. Timisoara.. 3. Numerical. Canada. 2001. Fluent Incorporated. 4. Champagne. Pinkas. . Analysis Of A Standard Pump In Reverse Operation Using CFD. REFERENCES 1. Proceeding of Flucome. The cavitation risk for the bigger flow rates is higher than for the smaller flow rates. Muntean. A. Graz. Stoffel.Y. FLUENT 6. Simulation of threedimensional unsteady flow in hydraulic pumps.. S. 20th IAHR Symposyum. Gambit 2. Ludwig. G... M. Thesis University of Twente.. Fluent Incorporated. B... Ludwig. Stoffel.. Experimental And Numerical Studies On A Centrifugal Pump With 2d-Curved Blades In Cavitating Condition. Proceeding of Fourth International Symposium on Cavitation CAV2001. A.. 2001. Numerical methods for the analysis of the 3D flow in Francis turbine runners. 1997. 1999 7.. 2001. Louisiana. Van Esch. Tamm.. M... 2000 2. P. May 29 – June 1. Experimental And Theoretical Analysis Of The Individual Efficiencies of a Centrifugal Pump. 2002. Hofmann.. Modern Pump Impeller Design With Consideration Of Non-Uniform Inlet Flow Fields. E. Sebestyen. Fluent Inc. The Internal Flow Investigation of a Centrifugal Pump. J.4. The numerical simulation predicts a minimum pressure on the suction side near the crown. Mannschreck. G. B. Hydraulic Machinery and Systems.. W. User’s Guide.. California. SUA 5. in agreement with experimental observations. ACKNOWLEDGMENTS The authors acknowledge the support from the National University Research Council grants. B. 9. Akhras. Fluent Inc. M.. A. Tamm. Stoffel. Bråten... PhD Thesis Politehnica University of Timisoara. A. User’s Guide. All numerical computations have been performed at the Numerical Simulation and Parallel Computing Laboratory of the “Politehnica” University of Timisoara. National Centre for Engineering of Systems with Complex Fluids. B. Morel. A.. 6. Enschede. 2001 8. Comparison between numerical results and data given by the pump producer is presented.

Figure 1 and a kl = 5 was used for NMV 2000RA. In [1] flows in a suction sump of a vertical pump. Bucuresti. Assist. were mentioned the dimensions and the geometry of a suction sump for a vertical pump.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). So the flow could not involve a large quantity of air. Prep. numerical simulation. suction sump.A. The suction sump for NMV 1000RA pump was physical modeled on a scale of kl = 2. Assist Department of Hydraulic Machinery “Politehnica” University of Bucharest Calin GHERGU. The physical model of the suction sumps level in the channel is fluctuating. the swirls types KEYWORDS and the protection elements to have a flow without Vertical pump. not at mixed flow pumps operation if the water level in the least. Department of Aerospace Sciences “Elie Carafoli” “Politehnica” University of Bucharest Georgiana DUNCA. Prof* Department of Hydraulic Machinery “Politehnica” University of Bucharest Marius STOIA-DJESKA. vortexes. by the working conditions of the pump. Would be a of the pumps is described for two types of vertical swirling flow at the lowest water level in the suction pumps: NMV 1000RA and NMV 2000RA. Prof Department of Hydraulic Machinery “Politehnica” University of Bucharest Diana Maria BUCUR. sump or not? To respond at this question two types Experimental results combined with numerical of vertical pumps were proposed: NMV 1000RA simulations give us the behavior of the swirling and NMV 2000RA and their sumps suction.C. Fax: (+40) 21 4029 523. Email: ecisbasoiu@yahoo. EXPERIMENTAL APPROACH Eugen Constantin ISBASOIU. experimental results. Department of Hydraulic Machinery “Politehnica” University of Bucharest Petrisor STANESCU.12. Assist. It will be obtained the The study on the physical model of the suction limit value of the suction head of installation pump conditions of a vertical pump deals with the inferior so as the pumps could have safety conditions of limit of the water level in the sump pump if the working. The experimental approach have the purpose to The swirling flow is influenced by the geometry determine the working conditions in axial and and the dimensions of the suction sump and. 2007 SWIRLING FLOWS IN THE SUCTION SUMPS OF VERTICAL PUMPS. Bucuresti Carmen Anca SAFTA. MEng. the intake channel lowers. [1]. Figure 2. INTRODUCTION Suction sumps are hydro-mechanical equipments used in the pump stations to receive water flowing from the intake channel to the pumps suction aria.com intake channel and the suction pump through the ABSTRACT sump. swirling flow. 060024. 1. [5]. Prof. Tests for NMV 1000RA and NMV 2000RA pumps were made. [3] or experimental using a hydraulic scale model [4]. Assist.2. Department of Hydraulic Machinery “Politehnica” University of Bucharest *Corresponding author: 313 Splaiul Independentei. The scale . 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. Romania Tel. Fascicola 3. S. The pump functionality to the designed parameters is looking for a flow without swirls between the 2. AVERSA S. velocity distribution.: (+40) 21 4029 523. PHYSICAL MODEL OF THE SUMP The flow study in the suction sump could be numerical [2]. Romania June 1 .

∆p is the pressure difference on the diaphragm. PARAMETERS MEASURED ON THE MODEL Classical methods were used to measure the working parameters. The flow rate. During these tests the velocity components were measured in two perpendicular sections and directions of the suction sump. was measured using the tighten section method by a diaphragm mounted on the discharge pipe of the pump. As a model pump was used a reference diameter of D = 400 mm. Q. The suction sump of the NMV 1000RA pump (the dimensions of the prototype) where D = 336 mm the discharge pipe of the pump. a is the difference of the level between the median manometer plane and the water level in the sump. ⎛d⎞ m = ⎜ ⎟ with d = 269 mm the diaphragm ⎝D⎠ diameter. and the suction head of the pump. Figure 3. In the case of axial and mixed flow vertical pumps. The pumping head or the head of the pump. TESTS RESULTS Five sets of measurements were done: two for the sump suction of NMV 1000RA pump and three for the sump of NMV 2000RA pump. kl. is defined as the difference between the discharge head of the pump. 2 Figure 2. for the particular case of equality between the speed triangle of the model pump and the real one. nq. The working measured parameters are in Table 1 with visual qualitative observations of the flow. H. Ha. the pumping head is done as p r V r2 H= + +a ρg 2 g (1) where pr is the pressure measured with the differential manometer placed on the out pipe of the pump. if it is considered as Figure 3 Measured parameters for axial and mixed flow pumps 4. are the results of the hydraulic similitude. Vr is the medium velocity of water in the discharge pipe of the pump.coefficients. α is the flow coefficient of the diaphragm. Hr. The suction sump of the NMV 2000RA pump (the dimensions of the prototype) 3. the origin the water level in the suction sump. ρ = 1000 kg/m3 water density at 200C temperature. of axial or diagonal pumps. So: Q= 2 ∆p πD 2 mα 4 ρ (2) Figure 1. flow rate and pressure. It could be seen in .

716 m. .25 0.629 0.546 0. Displaying the velocity components any disturbance of the velocity distribution from the turbulent profile indicates the presence of a parasite flow induced by a swirl flow.24 5.25 0.24 5. The velocities were measured with a Pitot-Prandtl Tube and a differential manometer.9 1.62 0.194 0.62 0.716 1.24 24. Velocity distribution on horizontal axis (1 and 2 measure section) and vertical axis (3 and 4 measure section).31 22. No submerged swirl [1] was formed at the entrance of the impeller. Hsb = 2. Table 1 H m 7.9 1.446 1.746 1.58 17. NMV 1000RA.44 0. without vortexes.87 Hsb m 2.646 0.625 3 Figure 6 and Figure 7 indicates a disturbance of the velocity components distribution and the swirling flows are imminent in these cases if the water level in the sump is changed (see Hsb in Table 1). is the static pressure at the impeller entrance. 3 and 4. Figure 4 and Figure 5 indicates no significant deviations from the turbulent velocity distribution in the sump of NMV 1000RA pump.87 17. ∆h No.586 0.04 0.4 Q m /s 0. The submerged head.8 5.628 0.42 0. 1 2 3 4 5 The pump NMV 1000RA mm 372 333 333 333 850 740 850 850 pr MPa 0.Figure 1 and Figure 2 the measure sections numbered with 1.946 1. Hsb.25 1. 2.4 1.996 Observations without vortex without vortex with vortex with vortex without vortex with vortex with vortex without vortex NMV 2000RA Figure 4.

Figure 5.646 m. Hsb = 0.716 m. with vortexes. Hsb = 2. . Velocity distribution on horizontal axis (1 and 2 measure section) and vertical axis (3 and 4 measure section). NMV 2000RA. NMV 2000RA.646 m. Figure 7. Hsb = 0. Velocity distribution on horizontal axis and vertical axis. without vortexes. NMV 1000RA. Figure 6. Velocity distribution on horizontal axis and vertical axis. with vortexes.

Figure 9 and Figure 10.A numerical simulation of the flow in the suction sump was continuing [1] in both cases of the sumps of NMV 1000RA and NMV 2000RA vertical pumps. Figure 8. At a lower water level in the suction sump of the pump a swirling flow was observed. x-z axis with 10 m3/s flow rate for NMV 2000RA pump Figure 9 Velocity distribution in the horizontal plane. NMV 2000RA pump . 10 m3/s flow rate. The NMV 2000RA pump was numerical modeled for the flow rates of 10 m3/s and 15m3/s. Figure 8 Velocity distribution in the vertical and median planes. In these cases the numerical results did not indicate a swirling flow that could generate problems in the pump operation. The FLUENT computer fluid dynamic code was using and the k-ε model in turbulent flow was applied. The NMV 1000RA pump was numerical modeled for the flow rates of 1 m3/s and 3 m3/s.

E. Munteanu T.746 m. Hydr. REFERENCES 1. pp 17-22. (2005) Optimization of approach flow conditions of vertical pumping systems by physical model investigation. 9th International Symposium on Transport Phenomena and dynamics of Rotating Machinery. by use of computational fluid dynamic (CFD) analysis. Swirling flows in sumps of vertical pumps. approach. J. Huston. Shibata T. Isbasoiu.617 m3/s. 5. FEDSM2005-77347.. • NMV 2000RA pump in the suction sump could work without vortexes if the submerged head Hsb is lower then 1. .. Minisci E. Effects of approach flow condition on pump sump design.. s... Eng. 4. FLUENT Inc.. Iwano R. Bul. TX. FEDSM2005-77330... Sci. (2005) Hydraulic design validation of the suction intake of a vertical centrifugal pump station. (2005).a.. Hawaii. of FEDSM2005.746 m to have a high flow rate.146 m and the flow rate is between 0.A. ASCE. Schafer F. Honolulu. (1996). It is no recommended in pump operation at submerged heads less then 0. CONCLUSIONS The paper regarding the experimental study of the swirling flows in the intake channel suction sump of vertical pumps. USA. 3. s.Figure 10 Pathlines in the suction sump with 15 m3/s flow rate for NMV 2000RA pump Politehnica of Timisoara. Huston. Cicatelli G. Hellmann D. Of 2. 5. Gustave A. Telib H. 1998. of FEDSM2005. Proc.. Trans on Mechanics.C. the suction Theoretical University 6. Numerical prediction of the submerged vortex and its application to the flow in pump sumps with and without a baffle plate. USA. 122(9). *** FLUENT 5 User’s Guide.586 m3/s and 0. Influence of the level upstream in the suction pump and the development of the swirling flow in the suction sump pump are observed. Proc. TX. The physical model and the results of the tests give us: • NMV 1000RA pump in the suction sump could work in good conditions without vortexes if the submerged head Hsb is more then 0.-H. pp 489-494... (2002). Safta C..a.

e. The former paper presented the experimental tests performed in the boundary layer wind tunnel called TASL-1 from the Wind Engineering Laboratory.23.axis dynamic amplitude Ay [Hz] x-axis frequency fx [Hz] y-axis frequency fy [s] x. a general power law profile with α=0.* Hydraulic and Environmental Protection Department.ro Ud.utcb. Romania June 1 – June 2.axis acceleration ay [m/s2] [m] x. Fax: +40 21 243 3660.axis dynamic amplitude Ax [m] y. E-mail: mircead@hidraulica. Romania.axis acceleration ax y. Technical University of Civil Engineering Bucharest *Corresponding author: Blvd. design and experimental implementation of the . sect.axis period Ty 1. Technical University of Civil Engineering Bucharest Costin Ioan COŞOIU.Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics 3rd Workshop on Vortex Dominated Flows Bucharest. 2007 SOME ASPECTS ABOUT A VORTEX GENERATING BUILDING MODEL PLACED UPWIND AN AEROELASTIC MODEL IN THE BOUNDARY LAYER WIND TUNNEL Mircea DEGERATU. Assuming different values of the free distance between the “obstacle building” and the studied one we tried to evaluate the structural influence of a building that is part of an urban agglomeration.Silviu Ştefan. Dynamic structural response NOMENCLATURE [m/s2] x. Assistant Hydraulic and Environmental Protection Department. Assistant Hydraulic and Environmental Protection Department. Prof. Technical University of Civil Engineering Bucharest Andrei GEORGESCU.y Using as reference the assumptions synthesized velocity in the paper “Dynamic wind tunnel tests for the gravity g [m/s2] Bucharest Tower Centre” we have tried to evaluate the influence of a vortex generating building model placed upwind of the aero elastic model of the tower. Prof. Lecturer Hydraulic and Environmental Protection Department. synthesized in this paper. the concept. Considering the actual construction development in Bucharest each aerodynamic study has to consider the influence of the built area that surrounds the studied building. INTRODUCTION The experimental studies performed. c).i. 2. The simulated atmospheric boundary layer is the same as the one used in the above mentioned paper. Technical University of Civil Engineering Bucharest Răzvan.: +40 21 243 3660.x [m/s] x-axis dynamic displacement ABSTRACT velocity [m/s] y-axis dynamic displacement Ud. 020396 Bucharest. Senior Lecturer Hydraulic and Environmental Protection Department.axis period Tx [s] y.Technical University of Civil Engineering Bucharest for a 107 m tall building. These experimental researches continue the topics of the paper “Dynamic wind tunnel tests for the Bucharest Tower Center” published in the Scientific Bulletin of the Politehnica University of Timisoara. Dynamic wind loads. We presented the following results: a). the similitude criteria. Hydraulic and Environmental Protection Department. KEYWORDS Wind engineering.Transactions on Mechanics. followed the determination of the vibration behavior of a tall structure with dynamic response exposed to the atmospheric boundary layer and include the influence of vortices generated by another upwind positioned building. b). Technical University of Civil Engineering Bucharest Lucian SANDU. Technical University of Civil Engineering Bucharest Liviu HAŞEGAN. Lacul Tei 124. the building structure. Tel.

We analyzed the influence of the vortices generated by the structure situated upwind of the studied model. The transposed results at the natural scale represent the vibration characteristics along the x and y axis of the structure at a reference height of 100 m above ground.ay).S-E wind. SE 95°-∞ and SW 5°-90. SE 95°-180. EXPERIMENTAL RESULTS OBTAINED USING THE MODEL The measurements were performed for two different values of the wind/ building incidence angle using a 1:100 scaled model (first campaign: θ=95° . These scales were obtained using certain limitations for the wind action on aero elastic structures (just as the studied building is in our case). periods (Tx. 180 cm. frequencies (fx. for eight different wind directions. The model of the structure situated upwind is 80 cm long. so we’ve indexed our results with two arguments (SE 95°-90. SE 95°-∞). This “obstacle building” was positioned in front of the model at three different distances: 90 cm. with no building upwards as shown in fig. the results obtained for the acceleration. SW 5°-∞).and the second campaign θ=5°S-W wind) using Reynolds numbers of the order of 105.y). 1a.Ud.c). 1c We determined the vibration response characteristics of the aero-elastic 1:100 scale model using both x and y directions with accelerometers as accelerations (ax. EXPERIMENTAL RESULTS TRANSPOSED TO THE REAL BUILDING The vibration characteristics obtained by the measurements performed on the 1:100 scale model for the six different situations were transposed at the real scale using the proper similitude scales for each parameter. SW 5°-∞). d). SE 95°-180. SW 5°. dynamic amplitudes of vibrations (Ax.The last ten diagrams present the vibration characteristics of the model due to loads induced by SW wind (SW 5°-90. e). 3. and ∞).The first ten diagrams present the vibration characteristics of the model due to loads induced by SE wind (SE 95°-90.180.180. 1b .x. These experiments were performed in order to observe the differences induced by the distance between the buildings as well as the wind simulated velocity profile. Ay).aero-elastic model of the building. Fig.b. Fig. frequency and dynamic amplitude both along wind and cross wind using as reference the building axis. the experimental conditions including atmospheric boundary layer simulation.Ty) and dynamic displacement velocities (Ud. 2. . 15 cm wide and 30 cm tall facing the incident air stream with the larger face crossing the wind direction in the experimental zone of the tunnel. The results are divided into two groups with respect to the wind/ building incidence angle (SE95° and SW-5°) each divided into three subgroups with respect to the position of the building situated upwind the model (90 cm. SW 5°. f). 1a Fig. 180 cm and infinite value (the reference value.fy). the measurement methodology. .

00 7.00 0.00 15.10 9.90 SV 5° .00 0.00 0.00 5.00 0.00 3.00 5.00 6.00 0.00 7.13 0.00 7.00 20.∞ Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 0.00 3.00 a y [m/s2] Prototip a y [m/s2] Model .00 9.00 500.12 130.00 0.00 5.00 2.90 SE 5° .00 3.00 7.? SV 5° .180 13.00 1.00 8.00 100.08 0.08 8.00 5.00 2.00 600.00 6.00 0.10 0.00 4.00 0.00 3.00 0.09 8.00 0.90 SV 5° .00 4.180 SV 5° .30 0.20 0.12 0.00 10.00 6.00 3.00 fx [Hz] Model 0.00 20.00 300.00 200.00 1.00 100.00 7.90 SV 5° .00 3.00 80.00 2.00 8.00 5.00 80.00 4.90 SV 5° .00 90.00 2.00 200.00 2.11 0.00 Ax [mm] Prototip 7.00 0.05 0 8.00 Re x 10 pe Prototip Re x 10-5 pe Model SV 5° .00 0.00 1.11 0.180 Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 -7 5.00 Re x 10 pe Prototip Re x 10-5 pe Model SV 5° .00 8.180 Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .25 0.00 a x [m/s2] Model 15.00 3.00 10.00 Tx [s] Prototip Ty [s] Model 0.05 0.15 0.00 100.00 0.00 600.00 0.08 0.00 4.12 fx [Hz] Prototip fy [Hz] Model 11.00 0.∞ SV 5° .? SV 5° .00 1.00 11.09 0.00 4.00 0.00 2.00 Ax [mm] Model 4.00 1.10 0.13 13.00 10.90 SV 5° .00 0.13 0.00 0.00 0.00 7.00 4.00 6.00 700.00 Ay [mm] Model 400.00 6.00 1.00 2.10 100.00 2.00 7.00 5.180 SV 5° .00 -7 5.00 1.00 3.00 12.00 6.25.00 500.? Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 4.00 3.? SV 5° .00 4.180 0.00 5.00 8.180 6.90 SV 5° .00 120.00 110.11 Tx [s] Model 110.00 7.15 0.25 0.00 3.1 0.00 300.00 0.00 2.00 fy [Hz] Prototip 12.00 Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 4.00 8.00 Re x 10 pe Prototip Re x 10-5 pe Model SV 5° .00 1.12 120.00 6.00 25.00 1.09 90.00 0.00 6.180 Series4 Ty [s] Prototip Ay [mm] Prototip 5.00 400.09 0.00 1.00 5.10 0.00 8.00 0.00 0.90 SV 5° .00 0.2 a x [m/s2] Prototip 30.00 2.? SV 5° .00 -7 5.00 6.? SV 5° .11 10.

00 2.00 6.00 8.1 0.15 0.90 SE 95° .09 0.00 4.11 0.180 9 8 7 Ax [mm] Model 6 5 4 3 2 1 0 0.00 6.00 2.12 0.00 2.00 Re x 10-7 pe Prototip Re x 10-5 pe Model SE 95° .09 1.90 SV 5° .15 0.00 8.25 0.00 4.00 7.25 0.00 3.00 1.30 0.00 400.10 0.00 Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 7.00 5.? SE 95° .5 1.00 600.180 12 11 fx [Hz] Model 10 9 8 7 6 0.3 0.? SV 5° .00 5.00 3.5 10 9.15 0.12 0.00 -7 0.10 0.25 0.15 0.00 500.00 -7 900.00 100.00 Re x 10 pe Prototip Re x 10-5 pe Model SE 95° .00 4.30 0.00 700.00 8.00 0.00 0.08 0.20 0.10 0.00 2.00 6.00 6.00 2.00 7.00 3.00 Re x 10 pe Prototip Re x 10-5 pe Model SE 95° .00 8 0.00 0.0.00 a y [m/s2] Prototip fy [Hz] Prototip Ay [mm] Prototip 20 15 10 5 0 0.00 6.00 4.00 2.05 0.00 SE 95° .00 100.20 0.00 200.00 3.00 Udy [m/s] Model 0.? SE 95° .00 3.05 0.05 0.00 7.00 4.12 0.00 Ay [mm] Model 7 6 700.? SV 5° .00 7.00 1.00 8.00 400.00 8.25 a x [m/s2] Prototip 0.00 0.180 Re x 10-7 pe Prototip Re x 10-5 pe Model SE 95° .07 0.10 0.? SE 95° .05 0.90 SE 95° .5 9 8.00 6.20 0.00 3.10 0.00 7.00 0.? SE 95° .00 6.00 2.00 6.15 0.00 5.10 0.00 7.00 4.35 0.00 800.00 5.00 7.00 Re x 10 pe Prototip Re x 10-5 pe Model SE 95° .00 -7 5.11 0.25 Udx [m/s] Model 0.05 0.00 5.05 0 5.11 0.90 SE 95° .10 0.00 Re x 10-7 pe Prototip Re x 10-5 pe Model 600.00 8.2 0.00 5.10 0.00 1.35 0.180 .00 1.00 200.25 Udx [m/s] Prototip 0.180 Ax [mm] Prototip 5 4 3 2 1 0 0.180 30 25 a x [m/s2] Model 20 15 10 5 0 0.06 8.180 Re x 10-7 pe Prototip Re x 10-5 pe Model SV 5° .00 4.20 0.30 0.00 0.90 SE 95° .00 1.00 0.00 2.00 10.00 fx [Hz] Prototip 12 11.00 3.90 SE 95° .00 500.00 300.00 1.09 0.08 8.90 SV 5° .20 0.00 300.00 0.00 3.180 30 25 a y [m/s2] Model 0.15 0.00 4.∞ SE 95° .30 0.5 11 fy [Hz] Model 0.? SE 95° .30 Udy [m/s] Prototip 0.90 SE 95° .

00 7.11 0.180 Series4 0.085 0. the vortices created upwind induce significant growth of the values referring to the x. Regarding the dynamic amplitudes..00 3.00 90.00 120.00 100.1 0 8.00 6.00 3.axis for the SW wind load and decrease the acceleration values on the y.00 105.? SE 95° .90 SE 95° .08 0. the vortices created upwards induce decreasing values reffering to both axis directions and both wind directions..09 0.00 4.2).5]* 107 because this domain offers significant values for the aeroelastic behavior of the analyzed structure.00 2.00 5. 2 .115 0. These numbers are big enough to insure the self simillarity of the longitudinal velocity profile for the inertial domain which includes the fluctuating components of the velocity that induce the resonance excitement for this tall building category. As the distance between the “obstacle building” decreases.15 0.180 4.axis for SW wind load..00 6.25 Udy [m/s] Model 0.90 SE 95° ..095 0.1 0. The present research confirms.00 85.00 1.2 0.00 6.00 140.5 . once again.00 2. Regarding the acceleration parameters.00 4.90 SE 95° .1 0.105 0. CONCLUSIONS The experimental studies were performed for Reynolds numbers in the range [2.axis for SE wind load and the x.00 7.00 80.2 0.00 2.00 8.25 0.00 0.4 Udx [m/s] Prototip 0.5 0.00 Udy [m/s] Prototip 0.5 .00 Re x 10-7 pe Prototip Re x 10-5 pe Model SE 95° .3 0..14 0.00 100.2 0. The final caption presents the model placed in the experimental vein of the aerodynamic wind tunnel (Fig.13 Tx [s] Model 0.08 0.00 80.1 0 0.00 8.12 0.05 0 0.00 6.5]* 107.00 Re x 10-7 pe Prototip Re x 10-5 pe Model SE 95° .1 0.1 0.00 4. The assumptions that will follow were made for Reynolds numbers in the range [6.2 0.180 SE 95° .05 0 8.00 0.3 0.0.5 0. 7.00 3.00 90.180 Re x 10 pe Prototip Re x 10-5 pe Model SE 95° .00 1.00 95.00 5.00 -7 5. 7.00 3.? SE 95° .00 Ty [s] Prototip 1.15 0.90 SE 95° .4 Udx [m/s] Model 0.3 0.00 110.∞ SE 95° .00 4.3 0.00 5. Fig.00 7.00 150.00 1.axis for SE wind load and the y.00 110.00 0. the necesity of performing this type of studies in order to get a good idea of the interaction between the atmospheric boundary layer and the slender structures..00 7.∞ Re x 10-7 pe Prototip Re x 10-5 pe Model SE 95° .00 120.09 0.15 0.12 0.00 2. the amplitude values decrease is stronger.11 Ty [s] Model 0.00 115.00 Tx [s] Prototip 130.

C. 1975. C. California. 8. Hasegan. C. L. L. 2332. on Mechanics.. (T21).. C.. Technical University of Civil Engineering. Georgescu.. 2005. 2. Davenport. L. Pascu. Georgescu. Iamandi. 2002. Modelling of wind action on the Bucharest Tower Center International Building (in Romanian). & Powell.REFERENCES 1. Coşoiu. Bull.. Romanian Standard Institute. Iamandi. Lubbock. 2003.. Stefan. L. 2002. Proc. Trans. Timişoara: Orizonturi Universitare. M..I. Bangkok. 2006 “Dynamic wind tunnel tests for the Bucharest tower center. Romania. Erbasu. 11th International Conference on Wind Engineering. EERC 75-37. Sandu. Bucharest.. 3. Georgescu. Degeratu. . Int. The application of the boundary layer wind tunnel to the prediction of wind loading.. Atmospheric boundary layer (in Romanian). M. Texas. A. Begell House Inc. Trans. C.P. 427/2004. Cosoiu... A.G. A... New York. 50 (64). L. D. Research report no. 7.. pp. Proc. A. Report No. A.”. “Atmospheric Boundary Layer Change”. 2005. Vol. of Fluid Mechanics Research. C. Haşegan. “Actual and Perspective Research in Wind Engineering”. Georgescu. Thailand.. 29. ISSN 1064-2277. 6. Iamandi.. Bull.. M. J. International Conference on Urban Engineering in Asian Cities. A. Degeratu. Politehnica University of Timisoara. 10.. “Experimental Modelling of a Four Steel Tanks Battery”. USA... 1996. Romania.. Vol. M. Sci. Georgescu. C. Sci. ANSR – I General Purpose Program for Analysis of Nonlinear Structural Response. Degeratu. Berkeley. M. 1968. Sandu. Degeratu. R. 4. STAS 10101/20. Degeratu. Hasegan. R. 1990. Sandu. on Mechanics. 9. A. Tom 51(65). 3. Georgescu.. Fasc. Toronto.. Politehnica University of Timisoara. L. of International Research Seminar of Wind Effect on Buildings and Structures. Mondkar. C. “Loads due to wind”. 5. Erbaşu. A wind tunnel study for a TV tower in Bucharest – Romania.H. G. Cosoiu.

and documentation of the CFD code. 800008. key issues include discretization of the continuous modeled PDEs.ro the following paragraphs. Email: adrian. a kinematic boundary condition is applied at the free-surface. Since the computational grid has to conform to the free-surface shape. development of numerical algorithms for solution of the discretized modeled equations. the CFD process is composed of two distinct parts. and testing the algorithm in a CFD code. KEYWORDS Free-surface flow. Romania Tel. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. (i) selection or development of a general-purpose CFD code and (ii) use of the CFD code for solution of a particular flow problem of interest. assessing risks in the use of CFD results in a design environment. [2]. K-ω SST turbulence model 1. LES. Galati. development of a comprehensive process is useful for training nonexpert CFD users. INTRODUCTION There is a significant effort in the marine industry to integrate the computational fluid dynamics (CFD) simulation capability in designing energy efficient ship hull forms while lowering the noise generated from them. establishing confidence in results from CFD codes.g. Fax: (+40) 236 495400. The flow problem to be simulated is rich in complexity and poses many modeling challenges because of the existence of breaking waves around the ship hull.2. DNS).: (+40) 236 495400. Specifically. As described in . Accurate simulation of turbulent free surface flows around surface ships has a central role in the optimal design of such naval vessels. 2007 FREE-SURFACE TURBULENT FLOW AROUND A LPG SHIP HULL Adrian LUNGU*. finite volume method.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). and because of the interaction of the two-phase flow with the turbulent boundary layer. and the governing equations are solved only for the water phase [1]. Various flow cases were studied in an attempt of finding the most suitable hull forms from the wave-resistance point of view. and streamlining the task of obtaining CFD solutions leading to reduced manpower requirements. and selection of a comprehensive set of initial and boundary conditions required to solve a wide range of applications. In interface tracking methods. Both approaches aim to compute the wave profile accurately because the wetted surface area appears in the calculation of the drag acting on ship motion. this approach is not efficient for high Froude number flows in which breaking ABSTRACT The paper presents a methodology for computing the 3D turbulent free-surface flow around a 10000 DwT LPG vessel. Professor Department of Ship Hydrodynamics “Dunarea de Jos” University of Galati *Corresponding author: 47 Domneasca Street. namely. Development of any general-purpose CFD code has several common elements. In general. formulation of the general initial and boundary value problem (IBVP) which is to be solved numerically using a CFD code. Key issues in the formulation of the IBVP are in definition of the scope and level of flow description (e. The flow is simulated by making use of a RANS-based finite volume approach. Closure to turbulence is attained through the K-ω SST turbulence model.. Although some of the elements of the CFD process are relatively straightforward. Fascicola 3. whereas the latter must be followed for each simulation. RANS. initial and boundary conditions. Overall philosophy for the CFD process is given in this section as a set of procedures to guide engineers and scientists through the process of modeling fluid flow problems using a CFD code. With regard to numerical methods. development of numerical methods for approximate solution of the IBVP. the former occurs only at infrequent intervals when need arises to make large shifts in technology. Romania June 1 . Generally speaking.lungu@ugal. selection of governing partial differential equations and physical models for the fluid flow. there are two distinct approaches in representing the free-surface in a Reynolds-averaged Navier-Stokes (RANS) simulation. interface tracking and interface capturing methods.

The existence of a free-surface interacting with the turbulent boundary layer presents challenges for turbulence modeling.waves with high amplitudes are typical. The previous k − ω models give somewhat more realistic solutions for such flows. the grid generation process for the current geometry proved to be a tedious one because of the existence of a bulbous fore. The characteristic length in Froude and Reynolds numbers is the distance measured between the perpendiculars of the ship. Volume of fluid (VOF) [3]. Although the gridless technique is continuously investigated. In addition. and gravity is effective on larger scales [6]. In 1993. is used outside of this region. such as poor implementation of the conservation laws. the turbulent boundary layer. such as a free-surface piercing hydrofoil. A fine grid around the interface is desirable to minimize the smearing of the interface due to numerical diffusion. The CFD technology is now putting more stress on customizing. transformed to resemble a k − ω model. may not adequately represent the turbulence structure at the free-surface. curvilinear structured grids. whereas the Reynolds number represents the ratio of inertial forces to the viscous forces. Existing turbulence models. it will be employed in the present study. The grid generation process is as follows. The k − ε models over-predict the turbulent shear stress in boundary layers under adverse pressure gradient. it is known that for turbulent breaking waves. one can choose either the rectangular grids or the boundary-fitted. it still has some substantial difficulties. The important dimensionless numbers defining the flow conditions for the free-surface flow around ships are the Froude and the Reynolds numbers. Presently we must work within the framework of the method with grid system. Fig. The spacing of the nearest grid cells are such that it . Furthermore. For instance. The k − ω SST model also has the advantage of being insensitive to the free stream ω value. A series of nine speeds ranging from 14 knots to 18 knots around the required speed is considered. which have been mainly proposed for single phase flows. With the k − ω SST model a significant improvement is reached. A hybrid structured mesh consisting of about 2⋅106 cells was generated for the present simulations.242 and a Reynolds number of 1. needs to be resolved with a mesh having a not too large aspect ratio. It is therefore necessary to assess carefully the performance of available turbulence models in literature for simulating turbulence flows with free-surface. The ship hull is supposed to perform longitudinal (trim) and vertical (sinkage) free movements in waves. The different sets of coefficients and the additional cross-diffusion term from the transformed k − ε model are combined by blending or switching functions Fi in an intermediate region. First. the simulation success relies on the grid generation to a larger extent than before.1 shows the overall grid structure adopted in the simulation. which correspond to a ship speed of 16 knots. respectively. 1.2⋅107 defines the standard flow conditions for the present case. the surface capturing methods are more versatile in handling a variety of free-surface flow conditions. Since we have reliable solution method for the N-S equations. Menter [7] published a new shear stress transport k − ω model which has remarkable advantages when compared with k − ε and previous k − ω models. The turbulent flow structure in the vicinity of the free-surface can also be much more complex than the turbulent flow structure of single phase flows. In the k − ω SST model the k − ω model is used near the wall and a k − ε model. The model is based on Bradshaw’s assumption that the principal shearstress is proportional to the turbulent kinetic energy. for which the grid generation technique plays an important role. which is introduced into the definition of the eddyviscosity. Actually a large part of the efforts is devoted to the generation when a 3D body of complex geometry is deal with. level-set [4] and front tracking [5] are the widely adopted techniques in this category. On the other hand. which can be seen in Fig. The Froude number is representative of the ratio of inertial forces to gravitational forces. the surface of the ship is described by an off-set file and then a boundary layer mesh is fitted all around it. Although the structured grids find difficulties in the representation of the boundary of extreme complexity. but the improvement is only partial. in which the governing equations are solved for both the air and the water phases. surface tension effects are important at smaller scales. 2. which is very thin due to the high Reynolds number. Since the unstructured grids like those for the finite element method still have difficulties in developing into 3D cases. Although these requirements can be met easily for a relatively simple geometry. GRID GENERATION In the past 30 years the technique of solving the Navier-Stokes (N-S hereafter) equations has made remarkable progress and the gradient of improvement has recently been decreased. A Froude number of 0. the VOF technique also benefits from adopting grid cells of hexahedral shape.

P = P + p '' ≡ p + p '' The time mean of a variable is defined as: (8) φ = lim 1 2 T →∞ T T −T ∫ φ ⋅ dt (9) The following rules of averaging apply for any two turbulent quantities φ1 and φ2: " " . The structured grid topology 3. In the present model. The computational domain is then meshed with quadrilateral shaped 2D cells. φ1 = φ1. then into 16 cells.1. mean pressure in time.1. φ1φ2 = φ1φ φ1''φ2 = 0. The rest of the terms are discretized with central differences. U i = U i + ui" ≡ ui + ui" (7) and the instant pressure. knowing the average will usually be enough. The convective terms are discretized with a Roe scheme and a second order explicit defect correction is used to approach second order accuracy. MATHEMATICAL MODEL The solver is a finite volume RANS (ReynoldsAveraged Navier-Stokes) one with a K-ω SST turbulence model. and time fluctuating pressure. which are the basis for the 3D hexahedral cells in the domain. φ1 + φ2 = φ1 + φ2 = ∂s ∂s (10 ) " " φ1φ2 = φ1φ2 + φ1 φ2 (1) Taking the time average of the continuity equation gives: Only incompressible flow is considered in the present study. which requires extremely dense grids to resolve the smallest turbulent length scales. 3. Solving the RANS equations will give the time average velocity and pressure. That means that the changes in density are negligible. p. φ1 φ1 = φ1 + φ1 = 0.g. A local artificial time-step is added to the equations and the discrete coupled equations are solved using an ADI-solver. each 3D cell is divided into 4 cells. but since the time fluctuating velocity and pressure in general are much smaller in amplitude. Then the continuity equation can be written: ∂U i ∂U i ∂ui = = =0 ∂xi ∂xi ∂xi (11) . the continuity equation states that mass is conserved: 1 ∂ρ ∂U i + =0 ρ ∂t ∂xi The RANS equations can be derived from (3) by splitting the instant velocity components Ui in time mean velocity ui. ∂φ1 ∂φ 1 . a local grid refinement (e. and time fluctuating velocity u″i. and so on) is applied around the free-surface and the hull extremities to provide reasonable resolution of the interface. Finally. Equations The N-S equations can be solved numerically by resolving all scales for turbulent flows. p″. ∂U i =0 ∂xi (2) The N-S equations of motion can be written in the following form: ρ ∂ U jU i ∂σ ij ∂U i = ρRi + +ρ ∂x j ∂x j ∂t ( ) (3) where σij is the total stress and for a Newtonian fluid can be written as: σ ij = − Pδ ij + 2µ ⎜ Sij − Skkδ ij ⎟ ⎛ ⎝ 1 3 ⎞ ⎠ (4) where Sij is the strain-rate defined as: 1 ⎛ ∂U ∂U j ⎞ ⎟ Sij = ⎜ i + 2⎜ ∂xi ⎟ ⎠ ⎝ ∂x j (5) Skk in (4) is zero for incompressible flow: Z X Y 1 ⎛ ∂U k ∂U k + S kk = ⎜ 2⎜ ∂xk ⎝ ∂xk ⎞ ∂U k ⎟ ⎟ = ∂x = 0 k ⎠ (6) Figure 1. P.is consistent with the wall function formulation.

the Reynolds-averaged equations can be written as follows: ∂ui ∂ u j ui 1 ∂p 2 ∂k + + − = Ri − ρ ∂xi 3 ∂xi ∂t ∂x j ⎛ ∂ ∂ ⎛ ⎜υ ⎜ ∂ui + u j + E⎜ ∂x j ⎜ ⎝ ∂x j ∂xi ⎝ ⎞⎞ ⎟⎟ ⎟⎟ ⎠⎠ − ρ Ri + ∂p ∂ − ∂xi ∂x j ⎛ ⎛ ∂u ⎜ µ ⎜ i + ∂u j ⎜ ⎜ ∂x j ∂xi ⎝ ⎝ ⎞⎞ ⎟⎟ ⎟⎟ ⎠⎠ ( ) (17) Hence. is created instead. νT = T ρ ρ (18) (13 ) + ∂ ∂x j ⎛ ⎛ ∂u ∂u j ⎜υ ⎜ i + ⎜ ⎜ ∂x j ∂xi ⎝ ⎝ ⎞⎞ ⎟⎟ ⎟⎟ ⎠⎠ The k equation can be derived by subtracting the NS equations for the mean flow (13) from the N-S equations for the instantaneous flow (3). The problem now is that the time averaging has created a tensor of unknowns. Therefore. In order to solve the system more equations are needed. the time averaged continuity equation and N-S equations for incompressible flow can be written as follows: ∂ui =0 ∂xi '' '' ∂⎛ ⎜ u j u i + u j ui ⎞ ⎟ ∂ui ⎝ ⎠ = R − 1 ∂p + + i ∂x j ρ ∂xi ∂t where νE is the effective kinematic viscosity defined as: ν E = ν +ν T . Additional equations can be formed by taking moments of the N-S equations. The last term in (15) is needed to get the proper trace of the Reynolds stress tensor ρ ui''u 'j' = − µT ⎜ ⎛ ∂ui ∂u j + ⎜ ∂x j ∂xi ⎝ ⎞ 2 ⎟ + ρk ⇒ ui''u 'j' = ⎟ 3 ⎠ =ρ ∂ U jU i ∂U i ∂ ∂P − +ρ − ρ Ri + ∂xi ∂x j ∂t ∂t ( ) (16 ) '' ' ' ⎞ ∂⎛ ⎜ u j ui + u j ui ⎟ ⎛ ⎛ ∂U ⎞⎞ ⎠− ⎜ µ ⎜ i + ∂U j ⎟ ⎟ = ρ ∂U i + ρ ⎝ ⎜ ⎜ ∂x j ∂xi ⎟ ⎟ ∂t ∂x j ⎠⎠ ⎝ ⎝ Thus. ν = µ µ . multiplying it with ui'' and taking the time average: ∂u ∂ ∂k ∂ u j k + = ui''u 'j' i − ε + ∂x j ∂x j ∂t ∂x j ( ) where: υ= µ ρ (14) ⎛ ∂k 1 '' '' '' 1 '' '' ⎞ ⎜ν − u u u − p u j ⎟. This is called the closure problem. This can be done by multiplying the N-S equations by a fluctuating property and time average the product. But these equations produce some new unknowns and solving all these equations takes a lot of CPU time. Closure of the k equation can be achieved by replacing the unknowns with approximations which are based on experimental data. k. the Boussinesq approximation for incompressible flows ρ ui''u 'j' = − µT ⎜ ⎛ ∂ui ∂u j + ⎜ ∂x j ∂xi ⎝ ⎞ 2 ⎟ + ρkδ ij ⎟ 3 ⎠ Then taking the time average of the N-S equations (first moving all the terms to the left hand side): ⎛ ∂ U jU i ∂U ∂P ∂ ⎛ ⎜ µ ⎜ ∂U i + ∂U j − ρRi + − ρ i +ρ ∂t ∂x j ∂xi ∂x j ⎜ ⎜ ∂x ∂xi ⎝ ⎝ j (15) ( ) ⎞⎞ ⎟⎟ ⎟⎟ ⎠⎠ is assumed valid and an equation for the turbulent kinetic energy. dissipation ( ε ) and turbulent diffusion (the last two terms on the right hand side).2. ⎜ ∂x j 2 i i j ρ ⎟ ⎝ ⎠ (19) 3.1. Turbulence modeling ε= It is through the Reynolds stresses − ρ ui''u 'j' that the turbulence makes their imprint upon the mean velocity and pressure. Since − ρ ui''u 'j' is symmetric there are six additional unknowns. This ∂ui''∂ui'' ∂xk ∂xk The equation for k is not closed since it contains unknown quantities for ρ ui''u 'j' . The last two terms in the first equation of the (19) set are usually combined and modeled as a gradientdiffusion term: .Subtracting (11) from (2) gives that also the time fluctuating velocity fulfills the incompressible continuity equation: ∂ui'' =0 ∂xi (12) procedure produces differential equations for the Reynolds stress tensor. The Reynolds stress tensor can be replaced by assuming that the Boussinesq approximation is valid.

Two are the key features that are taken into account. Cε 2 and σ ε are modeled closure coefficients. Table 1. The flow problem to be simulated is rich in complexity and poses many modeling challenges because of the existence of breaking waves around the ship hull. a challenge has always come from the uncertainty associated with the coexistent nonlinearities such as viscous interactions and freesurface tensions. RESULTS AND DISCUSSIONS As said above. Moving a blunt piercing body through the free surface often results in a bow wave that breaks either in a plunging or in a spilling manner.. The relationship between ω and ε is ω = ε 0. many difficulties arise when the air-water interface undergoes large deformations or becomes discontinuous. and because of the interaction of the two-phase flow with the turbulent boundary layer. Each element in the tri-diagonal matrix is a 6x6 matrix. This leads to quite a complex equation for ε containing considerably more double and triple correlations than the k equation. the aim of the present study was to investigate the flow features in order to determine the optimum shape of the hull form aimed at contributing the ship resistance. their speeds are considerably smaller than the phase speed. while the secondorder flux corrections are used as an explicit defect correction. 6. k ∂k =0 ∂ξ B ω = const. One of them is the k − ω model that in some ways performs better then the k − ε model and in some way worse.ν ∂k 1 1 − ui' 'ui' 'u 'j' − p ''u 'j' = T ρ σ k ∂x j 2 (20) ε still needs to be modeled and an equation for the turbulent dissipation can be created by taking 2ν ∂ui'' ∂ ∂x j ∂x j (21) contains the first-order Roe convective terms and the second order diffusive terms. BOUNDARY CONDITIONS The boundary conditions used are described in Table 1. ω ω = f (ut . since most solution techniques for waves modeling are valid only when the waves have sufficiently small slopes. however. and the particle accelerations are not large compared to the acceleration due to gravity.) ∂ω =0 ∂ξ B Outflow ∂ui =0 ∂ξ B p=0 ∂k =0 ∂ξ B ∂ω =0 ∂ξ B 5. some other turbulence models have been developed. The tri-diagonal systems that are solved . whereas the second is the flow structure around the stern. These correlations needs to be modeled and a far simpler equation is obtained ∂u ε ε2 ∂ε ∂ u j ε = −Cε 1 ui' 'u 'j' i − Cε 2 + + k k ∂x j ∂x j ∂t ∂ + ∂x j ⎛⎛ ⎞⎞ ⎜ ⎜ν + ν T ∂ε ⎟ ⎟ ⎟ ⎜⎜ σ ε ∂x j ⎟ ⎠⎠ ⎝⎝ ( ) (22 ) where Cε 1 . It is widely known that there is a significant effort in the marine industry to integrate the CFD simulation capability in designing energy efficient ship hull forms while lowering the noise generated from them. Accurate simulation of turbulent free surface flows around surface ships has a central role in the optimal design of such naval vessels. The first problem has relevance to the wave-making resistance problem. NUMERICAL SCHEME The ADI scheme is used to solve the flow equations. Studying the breaking bow wave has relevance for naval architects involved in the design of hull forms since as much as 15% of the resistance of a full-form ship hull is associated with its breaking bow wave. and applying it on the instantaneous N-S equations (3) and finally averaging in time everything.. For each sweep a local artificial time-step is calculated based on the CFL and von Neumann numbers in all directions except the implicit one. Boundary conditions formulation No-slip Slip Inflow u ui = 0 ui ni = 0 ui = const . and a fully non-linear free-surface boundary condition is required. In the implementation two layers of ghost cells are used.09k 4. Besides. as shown in Figure 2. ∂ui =0 ∂ξ B p ∂p ∂p ∂p =0 =0 =0 ∂ξ B ∂ξ B ∂ξ B k =0 k = const. The subject is difficult. The first is the front crest of the wave that develops around the ship stem. Since the k − ε model often does not produce satisfactory results. Moreover.

If. etc. Towing tank measurements have proven that hulls with a significant geometric gradient (i. which are topologically rather similar to the horseshoe vortices generated around a juncture.Figure 2. As it may be seen in . The separated flow forms several vortices around the body. Usually the bilge vortex and the propeller jet are nearly coaxial. contributing to the vorticity production in the immediate vicinity of the breaking wave-crest. These vortices result from the wave-induced separation due to the adverse pressure gradient in front of the body.al. in addition. up to some extent. The flow is completely three-dimensional and becomes more complicated as the vortices interact with the water wave and the boundary layer developed on the body surface. being able not only to discuss the flow conditions and uncertainty analysis. [10]. Figure 3. et. the estimation of the wave resistance under the breaking assumption is hampered by the uncertainty associated with the complexity of the phenomenon. see [10] and [11]. such as the flow around appendages. low curvature) may be subjected of a high viscous resistance. Free-surface profile around the ship stem An important issue of any free-surface flow under the breaking assumption is the separation that is exclusively provoked by the wave-induced effects. to those generated by the contingence between a separated flow around a juncture and the corresponding boundary layer. Muzaferija and Peric. µT . studied more extensively the phenomenon. the discussion in conjunction with Figure 4 proposes an analysis of the particle paths plotted in the field around the aft part of the hull. In spite of its engineering importance. blunt bodies with small bilge radius may determine strong vortical flow structures. the modification of the lines plan should be done interactively according to the numerical solution of the flow problem. Such a completely three-dimensional viscous flow often leads to a non-uniform velocity field in the propeller. The heterogeneous character of the flow is due to the number of vortices originating upstream of the piercing body. Friction coefficient around the hull The other area subjected to careful insight of the flow features regards the stern. That is. Although some studies have been performed either theoretically or through conventional tank tests. Improving the lines plan in the fore part of the ship usually requires trial computations. Figure 3 shows the friction coefficient field computed around the hull at a speed of 16 Knot. The figure also holds several transverse cuts in the fields of the turbulence quantity. This is because a poor definition of the forms may easily produce a strong wake. When an oncoming free-surface flow encounters a piercing body. complex free-surface deformations occur as a result of the breaking inception. there is no established method for estimating the wave resistance or wake characteristics because the detailed flow mechanism is not completely understood. The phenomenon was firstly identified by Tahara and Stern [8]. they are co-rotating an accumulation effect will occur. visualizations and comparisons of the friction coefficient field around the ship shoulders for all the considered speeds. as well as the topology of the corresponding streamlines. Since areas with such a curvature should be avoided in the design process. Later. the flow may separate as a result of a sort of blocking effect. Apart of the three dimensional separation that eventually takes place. which eventually affects the propulsion performances. Complex interactions that produce there unveil topological flow structures that resemble. accompanied by separations and reattachments. In this respect.e. antennas or periscopes mounted on submarine fuselages. but also to derive and verify a new topological rule for a surface-piercing body configuration. [9] as well as Olivieri. a fact that should be avoided for cavitational reasons. A. This convoluted three-dimensional flow can be found in many places where a piercing body intersects the water surface.

this is seldom the case in the hull wake. Although not ideal. A propeller in a uniform onset flow tends to accelerate the flow ahead as well as in the jet behind. . is carried with the flow and is spread by diffusion. The enhancement of the vortex strength by stretching effects is also responsible for the acceleration of the flow in the central region near the free surface. A nodal point of separation is placed in both sides of the symmetry plane. Figure 5. the flow is completely threedimensional. two minima away from the symmetry plane but at or near the free surface are commonly observed. on the understanding that at a high Reynolds number the convection dominates the diffusion.the proposed figure. as Figure 6 bears out. As bilge vortices in a nominal wake are effective in reducing the wake peak in the top sector of the propeller disk. Apparently the viscous diffusion effects are so strong as to more than neutralize the acceleration effect. Figure 4. the stronger vortices in the propelled-hull wake have a proportional effect on the flow in the central region near the free surface. In this respect. The wake of a ship with running propeller is quite different from the wake of a towed hull (nominal wake). A coalescent line to which all the streamlines converge is extended over about a half of the draught.5 depicts the vorticity iso-surfaces around the ship stern. For a qualitative analysis of the interaction between the propeller and the nominal wake. one must recall that vorticity. The vorticity iso-surfaces plotted around the ship stern Where in most towed-hull wakes the smallest axial velocity occurs in the symmetry plane. and to weakly decelerate the flow outside these regions. the horizontal projections of the streamlines being substantially different from an xplane to the other. Fig. The differences are not only related to the diminished momentum deficit but also to the interaction between the propeller-induced flow and the hull wake. once generated. it is worth mentioning that such a flow structure represents the best that we could get after a big deal of numerical computations. Streamlines topology computed around the ship stern The topological structure of the streamlines inside the boundary layer is characterized by the appearance of two pairs of divergence-coalescence lines just above the bulb stern. Ahead of the propeller and outside the jet there is no propellerinduced rotation.

8. Fluids. sinkage and trim. E. Proceedings of the Gothenburg 2000.. (2001) The dynamics of strong turbulence at free surfaces Part 1: Description. Journal of Computational Physics 1996. Stern. The axial velocity iso-surfaces plotted around the ship stern 7. Li. Osher. (1998) A two fluid Navier-Stokes solver to simulate water entry. J.. 5.. T. Harlow. L. Comp. J. Meth. F. (1981) Volume of fluid method for the dynamics of free boundaries. S. 421. (1965) Numerical calculation of time-dependent viscous incompressible flows of fluid with free surface. Phys. & Nichols. 3. Avanzini. pp. 4. J. Menter.. & Sethian. 7. D. Olivieri. S. Phys. 201. (2001) Unsteady RANS method for surface ship boundary layer and wake and wake field. Meth. F. In Proceedings of 22nd Symposium on Naval Hydrodynamics. F. J. B. R. & Penna. (1993) Zonal Two Equation k − ω Turbulence Models for Aerodynamic Flows. J. F. boundary layer. Fluids 8. IIHR Technical Report No. Fluid Mech. 11. Chalmers University of Technology. Washington DC. Stern. 39. Sweden. & Stern. H.Figure 6. 6. Hirt. Muzaferija. Comp. 37. R. 445. Matusiak. & Welch. C. 9. M. 10. H. A. Larsson. A boundary fitted co-ordinate system is employed to allow an accurate formulation of the boundary conditions. The University of Iowa. Num. A large-domain approach for calculating ship boundary layers and wakes and wave fields for nonzero Froude number. W. ACKNOWLEDGMENTS The reported research could not be possible without the generosity of the Romanian National University Research Council (CNCSIS) which is greatly acknowledged for the provided financial support through the Grant A_C code 33/2005. S. J. 601. Fluids. & Peregrine. J. 2. Int. The numerical method proves the expected robustness in dealing with impact wave problems. 79. (2000) Numerical simulation of viscous flows with free surface around realistic hull forms with transom. (1988) Fronts propagating with curvature dependent speed: Algorithms based on Hamilton Jacobi formulations. Bertram. D. (2001) Towing tank experiments of resistance. 2182. 24th Fluid Dynamics Conference. Int.R. (2000) A workshop on numerical ship hydrodynamics. 127: 348. CONCLUSIONS The paper presents a methodology for computing the 3D incompressible turbulent free-surface flow around a 10000 DwT LNG carrier. Rhee. Num. Third-order upwind schemes are used to determine the location of the water particles laying on the free surface for the sake of speeding up the wave elevation. 37. Pistani. and free surface flow around a naval combatant INSEAN 2340 model.. M. AIAA paper-93-2906. . H. J. Stern F. Gothenburg. REFERENCES 1. F. Tahara Y. F. 449. 12. Brocchini.225-238. & Lehtimaki. V.. & Peric. Phys. A.

method of manufactured solutions and benchmark solutions for verification and use of experimental data for validation) defining V&V. Opportunities and challenges for achieving consensus and standard validation and certification methodology and procedures are discussed. providing a quantitative metric for verification. 2007 UNCERTAINTIES IN THE FREE-SURFACE POTENTIAL FLOW CODE SOLUTION Adrian LUNGU*. Quantitative metrics are needed for both verification and validation. many case studies demonstrating use of the grid convergence index. Such definitions.. A similar situation existed for experimental fluid dynamics (EFD) uncertainty analysis ca.lungu@ugal. benchmark cases. including some findings about the uncertainty estimates. but also on the water free surface. quantify. Professor Department of Ship Hydrodynamics “Dunarea de Jos” University of Galati *Corresponding author: 47 Domneasca Street. Code verification activities measure accuracy in relation to benchmark analytical and ordinary and partial differential equation solutions for simplified problems along with software quality assurance: identify. and methodology is needed for combining errors and uncertainties. including: use of definitions for errors and uncertainties. Galati.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). Rig- ABSTRACT The paper proposes a numerical investigation based on the boundary element method for solving the free-surface flow around the ship hull. Model validation activities measure accuracy in relation to experimental data with emphasis on validation tiers based on unit problems. and complete systems: identify and quantify error and uncertainty in the conceptual and computational models. numerical issues discretization 1. Pioneering work was done by Roache [1] who proposed the grid convergence index for estimating uncertainty due to grid and time step errors based on Richardson extrapolation using multiple solutions on systematically refined grids. discussion for validation of both codes and solutions. discusses mostly code.ro verification deals with mathematics and validation with physics) and activities (e. there are currently many viewpoints covering all aspects from basic concepts and definitions to detailed methodology and procedures..: (+40) 236 495400. although widespread use is still lacking. and broader issues such as code quality assurance and certification. and reduce errors in the computational model and its numerical solution. verification deals with equations solved correctly and validation with correct equations and .2. Romania June 1 . Email: adrian. [2] expanded on this work through overall discussion of verification and validation. however. convergence tests. 1960 for which currently standards are widely accepted and available.g. follows Roache phrases and expands considerably on his activities along with broad statements in defining validation. in the author viewpoint are inadequate. quantify the numerical error in the computational solution. Solution verification largely follows Roache along with consistency and iterative convergence checks. The present paper summarizes the verification and validation (V&V) of a CFD simulation by analyzing the SHIPFLOW CFD code solutions. Fax: (+40) 236 495400. and compare the computational and experimental results. Ref. Romania Tel. phrases (e. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. thereby. but also solutions and additionally discusses policy statements on experimental and numerical accuracy.g. Non-linear boundary condition is imposed not only on the hull. 800008. subsystem cases. single grid error estimation methods. estimate the experimental uncertainty. The AIAA Committee on Standards for CFD [3] and Guide for V&V of CFD Simulations [4] uses definitions from information theory for errors and uncertainties with emphasis on measurement of accuracy as opposed to estimation of errors and uncertainties. Fascicola 3. INTRODUCTION In spite of the ever-increasing need and importance for standards for computational fluid dynamics (CFD) uncertainty analysis/accuracy estimation. KEYWORDS Free-surface potential flow.

1 shows. if appropriate. initial. and (2) the concept of correction factors based on analytical benchmarks. The momentum integral equations are solved along streamlines traced from the potential flow solution. which is used to obtain a corrected solution. 2. which mostly focus on verification procedures for 2D problems. which differs considerably from previous approaches. The reason for the division of the flow field into zones is that the computational time may be reduced considerably compared to the global approach where the Navier-Stokes method is used in the entire computational domain. The numerical investigation . boundary and other conditions. and numerous case studies. including use of EFD definitions for errors and uncertainties. as Fig. A free-surface potentialflow method of Rankine-source type is used. The approach combines previous V&V approach with extensions of concept of N-version testing for consideration bias uncertainties and use of reference values (experimental data and uncertainties) for estimating interval of certification. including some findings about the uncertainty estimates. The present paper summarizes and combines the systematic check and verification of the SHIPFLOW CFD code approaches. which is considered essential in the author’s viewpoint similarly as it is an essential and integral part of EFD uncertainty analysis. The methodology is based on concepts. Previously developed validation methodology and procedures for estimating modeling errors and uncertainties [11] were extended to include the option of use of corrected solutions. the third zone includes the aft part of the hull and extends about half a ship length downstream of the hull. The authors of [9] developed an alternative quantitative approach to solution validation specifically for already developed CFD codes for industrial applications (geometry and domain. The second zone is a thin layer at the hull surface and a boundary layer method of the momentum integral type is used. 6]. In general this literature follows approaches similar to that described above.orous implementation is impressive [5. nonetheless. and equations derived for simulation errors and uncertainties. The zones are computed in sequence and boundary conditions are generated for succeeding zones. which in combination provides N-order-replication level and increased confidence for EFD uncertainty analysis. Certification is defined as a process for assessing probabilistic confidence intervals for CFD codes for specific benchmark applications and certification variables. The literature also includes editorial policy statements [7]. The philosophy is strongly influenced by EFD uncertainty analysis [10]. It also covers about half a ship length in the radial direction. 3. RESULTS AND DISCUSSIONS Benefiting of a powerful computational tool. As with V&V there are many viewpoints on certification [2]. range of applications requires interpolation and extrapolation methods. The panels were distributed on the hull. Similarly. The validation approach [9] has been shown successful in establishing intervals of V&V for RANS simulations for ship hydrodynamics through its use at recent Gothenburg 2000 Workshop on CFD in Ship Hydrodynamics [12-19]. Presumably. additional guidelines [8]. A NavierStokes method of the RANS type using the k–ε model and a wall-law is used in zone three. which provide the overall mathematical framework. fluid properties) and input parameters (such as iteration numbers and grid and time step sizes). Another problem with such definitions is lack of an overall mathematical framework for V&V. Finally. The flow domain is divided into three zones. and its uncertainty. it is assumed that model validation for simplified problems has also already been dealt with during model development. Opportunities and challenges for achieving consensus and standard V&V and certification methodology and procedures are discussed. A shortcoming of this and other V&V approaches is that the justification for uncertainty estimates at 95% confidence level is based on reasoning similar to that used for EFD bias uncertainties at the 0-order-replication level without additional statistical 1-order-replication level precision uncertainties. according to the lines plan provided by ICEPRONAV Galati. It is assumed that code verification and quality assurance issues have already been dealt with during code development. Verification procedures for estimating numerical errors and uncertainties include (1) the options of estimating the numerical uncertainty or the numerical error itself. systematic computations were performed on a 10000 DwT LPG ship to determine the optimal hull forms from the resistance reduction point of view. The first zone covers the entire hull and a part of its surrounding free-surface. these definitions are subject to same criticisms mentioned earlier. GENERAL FEATURES OF THE SOLVER A zonal approach is used in SHIPFLOW to compute the flow around the ship hull. models. and a computational method is developed for each zone. The computational method is based on a hybrid BEMRANSE method that employs the zonal approach. definitions.

25 0.0 1.5 2. Taking into account that to avoid the unwanted numerical interference between the hull and outflow boundary conditions induced by the Sommerfeld radiation condition (although Shipflow uses the Orlanski scheme) it is necessary to have a sufficiently extended domain at the downstream to allow the development of minimum two wave crests. Although five speeds are usually enough to determine the ship resistance. Table 1. the second being truncated around its half.0 1.0 1.7 0.0 1.7 0. numerical and geometrical parameters.8 0.5 2. which represents the required nominal speed of the ship.7 0.5 1.has been aimed at clarifying the influences that the numerical parameters may have on the accuracy of the computed solution.0 1.0 1.0 1.7 0. Z Y X pu tat io Computational parameter x/Lpp upstream Hull stations 48 48 48 48 48 48 48 48 48 48 48 48 48 48 48 48 30 35 40 45 49 50 52 55 48 x/Lpp downstream Speed [knot] 16 16 16 14-18 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 14-18 y/Lpp 1. For this reasons all the following computations will be performed on a domain extended downstream till x/Lpp=2.75 2.7 0. First.5 Lpp and 3.6 0.5 2.5 Lpp (computational cases No.5 2.5 2.7 0.5 2.7 0.5 2.5 0.7 0.0 1. Computational parameters for the flow simulation 1 2 3 4-12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33-41 1.5. To help the reader get a complete understanding of the aforementioned dependencies. Panel distribution on the hull Knowing the importance of the computational domain size on the overall accuracy of the numerical solution. 41 computational cases were carried out for nine different speeds within a range of 4 knots between 14 and 18.7 0.7 0. The theory of the numerical simulation of the free-surface flow suggests a minimum of 30 panels over a wave length as well as a minimum of two significant waves to be accommodated behind the hull.0 1.6 0.5 2. whereas for the largest downstream domain.5 2. the influence of the computational domain size at the downstream was studied.0 1.7 0. the research has been extended to a larger amount of pilot computations for the sake of clarifying completely the dependence between the flow characteristics and the physical.9 1.7 0.7 0.5 2.0 1. 13 and 14). measured in respect to the fore perpendicular.5 2.5 2. Figs.7 0.7 0.0 3.5 2.5Lpp and 3.5 2.0 1.5 2.5 2. All the simulations were performed at 16 Knot.7 0.0 1.5 0.5 2.0 1.5 2.0 1.7 0.5 0. Under these circumstances. 2 and 3 depict comparisons between the free-surface topologies computed for the domains with 1.0 0. we considered six different domain sizes at the downstream ranging between 5 Lpp and 3.0 1.7 In the case of the smallest x/Lpp.5 3.0 1.7 0.0 2. the real size of the downstream domain is smaller with a ship length. which were successively modified. Since the domain length is expressed from the fore perpendicular.5 2.0 1.7 0.5 Lpp respectively 2. as stated before. 8. the computational domain accommodates a wave only.0 Z Vedere din prova X Y Vedere din pupa Z X Y Figure 1. 1…3. the features of all the computational cases are tabulated in Table 1.5 Lpp dimensions measured from the ship fore perpendicular. as the figures mentioned above show. we considered as appropriate to perform systematic computations aimed at clarifying the influence of each numerical parameter in the simulation process.3 0.4 0.7 0. the maximum wave numbers is seven. .

25Lpp and 0.0 2. As it may be seen in Fig.5Lpp at downstream Fig. The computed values of the wave coefficient are tabulated in Table 2 for the six cases discussed above. This fact is explained by a possible dissipation produced by the larger panel size at the downstream.30 -0..99% +12.5 x/Lpp 3 0.5 is analyzed.573202 Cw×10 Cw variation +7. reveals a small phase shift.5 3. Variation of the wave coefficient with the downstream size of the computational domain 1.5Lpp domain. It is important to observe that when the downstream extension decreases from 2. Since the computational domain extends from 1.579638 0.4 0.50 x/L downstream= 1.95% +9.55% 0 -0. (computational case 14).50 x/L downstream= 3.4. Comparison between the free-surface topologies computed on domains of 1.50 0.5Lpp. From the tabulated data it may be seen that the wave resistance coefficient decreases with the downstream domain size increase.620592 0.5Lpp Table 2.25 -0..5Lpp (which corresponds to case 14). whereas the downstream size increases form 2.47% Table 3. Same conclusion applies to the second crest as well as to the following pseudo-crest.595788 0.5Lpp and 3.637831 0. which presents a comparison between the wave profiles computed on the on the sip hull for all the upstream dimension between 0. the decrease Cw is only 1. knowing that a rougher paneling induces a numerical damping of the computed solution.5Lpp to 3.5Lpp.41% +0.5Lpp in the six computational cases (1.597301 Cw×103 Cw variation +17.005 0. the numerical solutions are reported to that computed for the domain of 2.5 1. Cw increases by 7.660760 0.690257 0.9.47%. Variation of the wave coefficient with the upstream size of the computational domain -0. the wave computed for the 3.5Lpp and 3.603862 0.75 2.60 x/Lpp 0.5Lpp at downstream Figure 3.5Lpp in size at the downstream.70 0. This fact may lead to the conclusion that the difference between the computed wave resistances is not significant either. Although the variation of the domain size is significant. the magnitude of the first wave crest increases with the decrease of the upstream domain.0 3.2 0. because if Fig.03% +3.5Lpp (corresponding to the computational case 1) to 3.37% -1. 0.6% +2. the wetted hull surface profiles are not revealing major differences.585007 0.75 x/L downstream= 2.5Lpp to 1.00 x/L downstream= 3.005 -0. 13 and 14) discussed.010 0.40 -0.015 h/L x/L downstream= 1.6 0.010 0. it may easily seen that thing change significantly behind the hull.4 shows a comparison between the wave cuts drawn on the ship hull for a domain extending at the downstream between 1.22% +2.0 Figure 4.5Lpp to 3.11% -0.00 x/L downstream= 2. The situation changes when the wave cuts on the ship hull are considered. Moreover.8 x/L 1.7Lpp.585007 0 .0 0.000 -0. Comparison between the free-surface topologies computed on domains of 2.5Lpp and 3.6%.Figure 2. This is valid only for the waves profiles on the hull.581786 0. where an important drop of the wave magnitude takes place as the computational domain increases in length. Comparison between the wave cuts on the hull computed on domains extending at the downstream from 1.50 -0.

75 x/L downstream= 2.8.50 x/L downstream= 1.0 1..25Lpp the difference between the wave coefficient and that corresponding computed for 0. As expected.38% -1.50 x/L downstream= 3. Six different computational domains extended in front of the ship from 0. measured from the fore perpendicular.88% -3.5329403 0.Based on this finding. respectively of 0.7Lpp at the upstream Figure 7.0 x/L Figure 5.25Lpp and 0.573871 Cw×103 Cw variation -8.0Lpp are considered (computational cases 20.7% .5Lpp In the followings. we again came up to the decision that all the following computations in the present study be performed on a computational domain of a size at the downstream of 2.0 3.25Lpp and 0.80 0.010 0. Next.25Lpp and 0.60 0.24 and again 8 in Table 1).7Lpp.19 and 8 in Table 1) are considered.90% 50 0. Figures 6 and 7 show comparisons between the free-surface topologies computed for the domain sizes of 0.5230703 0. which shows a comparison between the wave profiles computed for all the six upstream dimensions of the free-surface.5568385 Cw×103 Cw variation +9.550575 0.96% +10.5Lpp and 0. is almost 18%.50 0. measured form the fore perpendicular.7Lpp. the influence of the lateral size of the computational domain on the wave coefficient is checked. 0. Comparison between the free-surface topologies computed on domains of 0.5850070 0 55 0. It is important to point out that in the case of 0.84% +13.7Lpp at the upstream 1.00 x/L downstream= 2.005 0.. Comparison between the free-surface topologies computed on domains of 0.010 domain.7Lpp (computational cases 15. Same conclusion may be withdrawn based on the analysis of Fig. Data in Table 3 prove that with the decrease of the upstream Figure 6.5Lpp and 3. 0. Based on this finding all the following computations will be carried out for a domain having a size at the upstream equal to 0.5Lpp. the two figures prove that the variation of the domain size at the upstream does not determine significant modifications of the wave pattern at the downstream.7Lpp measured from the ship stem. Variation of the wave coefficient with the lateral size of the computational domain 0. Comparison between the wave profiles computed on a domain having the downstream size between 1.00 x/L downstream= 3.7Lpp.25Lpp and 0.000 -0.06% Table 5. the influence of the free surface extension at the upstream is studied.015 h/L x/L downstream= 1.5154429 0. a value that corresponds to the 0. For larger sizes of the upstream the gap starts to decrease to 2% approximately.0 2. All the simulations are performed for the nominal speed of the ship.565203 0.7Lpp.74% +5.90 y/Lpp 0.31% -5.5Lpp and 0.00 0.589905 +0.70 0. ranging from 0.50 0.5Lpp and 1. All the simulations were performed for the nominal speed. the wave coefficient increases significantly. of stations 30 35 40 45 0.005 -0. Variation of the wave coefficient with the panel resolution No.77% +11. Six domain breadths varying from 0.536393 0.589124 0..7Lpp case.83% Table 4. A positive aspect is that regardless the domain size at the upstream the wave train does not modify the phase.5282885 0..

6 0.2124 h/L x/L upstream = -0.2 1.010 0.015 -0.60 x/L upstream= -0.0Lpp in breadth. it is worth mentioning that modifying laterally the domain size does not lead to a phase shift or to a wave-length modification.80 y/L = 0. Nine computational cases are proposed to which corresponds a hull discretization ranging from 33 to 55 stations (cases 25. the following simulations are performed on a domain of 1. 0.50 y/L = 0. then increasing to 14% h/L y/L = 0.2 0.00 -0.005 -0.7Lpp which determines the smallest wave resistance coefficient represents the default value of the Shipflow.005 -0. Again.8 x/L 1.3312 0.4 0.7Lpp case.25Lpp to 0.2 0. Comparison between the wave cuts on the hull computed on domains extending at the upstream from 0.5845 Cw×103 0. Comparison between the wave cuts on the hull computed on domains extending laterally from 0.010 0.6255 17 0.015 -0.4 x/L 2.4 0.2 0.0 0.0 2.25Lpp and 0.010 -0. For wider domains. a value that corresponds to the 0.30 x/L upstream= -0.5 0.005 0.7Lpp 0.000 Figure 10..60 y/L = 0.4 1.5 16 0.70 y/L = 0. It is worth mentioning that the size of 0.60 y/L = 0.8 2. Comparison between the wave profiles computed on a domain having the upstream size between 0.0 2.40 x/L upstream= -0. Based on this finding.9Lpp case.25 x/L upstream= -0.8 -0.005 0.0 0.005 0. a fact that was somewhat expectable to happen.6 0.010 0.50 y/L = 0.8 x/L 1.329 0.0 1. in Table 1).0 x/L 0. It is important to point out that the difference between the computed C w for the 0.30 x/L upstream= -0.4513 0.90 y/L = 1.0Lpp In the following.005 0.015 16. the variation of the lateral dimension of the domain brings forth significant changes of the wave patterns at the downstream. Variation of the wave coefficient with the ship speed Speed 14 14.0Lpp case is of about 10%.8 1.000 h/L x/L upstream= -0.010 0.010 -0.0 Figure 11.005 0.50 x/L upstream= -0. As it may be seen in the aforementioned figures. the value of this gap decreases gradually to about 5%.0Lpp .5 15 15.6 -0.Table 6.5673 0. All the simulations were performed for the nominal ship speed.70 y/L = 0.7Lpp Figures 10 and 11 present comparisons between the wave profiles computed in all the six lateral domain sizes.005 h/L y/L = 0.40 x/L upstream= -0.000 -0. Figures in Table 4 prove that with the decrease of the lateral size of the domain a decrease of the wave coefficient is obtained.80 y/L = 0..32 and 8 respectively.2 2. Comparison between the wave profiles computed on a domain having the lateral size between 0.6 1.70 -0.0 Figure 9.010 0.000 approximately.70 0. 48 panels correspond to the case 8.2 0.90 y/L = 1.00 0. which corresponds to the 0.4 -0.8382 18 1.5Lpp to 1.60 x/L upstream= -0.0 0.0 1.6 0.5 0.015 0.5Lpp and 1.25 x/L upstream= -0.50 x/L upstream= -0.010 Figure 8.5Lpp case and that corresponds to the 1.4 0.7192 17.6 0. 0. a study on the influence of the panel number on the flow solution is proposed.

55 de stations The free-surface flow theory prescribes the wave length as being: λ = 2πLPP Fn 2 . therefore in the areas where the hydrodynamic parameters are expected to display important gradients.. then changing the sign till 0. which depicts a comparison between the wave profiles computed in computational cases 4. In the case of a discretization with 30 panels. in case 8 (Fig. Comparison between the wave profiles computed for hull discretization with 30.000 -0. one may see the good agreement between the computed freesurface and that recommended by the wave theory. The aim of these simulations is to establish the wave coefficient variation. therefore for the three speeds mentioned above. Figure 12. the wave coefficient decreases as well.5 knots.0 1. Data in Table 5.005 -0. 16 and 18 knot respectively.0 0. Fig.000 -0. 12 bears comparisons between the freesurface topologies and between the pressure fields on the fore ship extremity computed for 30 and 48 panels over the ship hull.0 Figure 13. The aim of such purpose is to investigate the influence on the numerical solution of the panel clustering at the ship extremities.83% for 55 panels on the hull. the difference between the wave coefficients is more than 8%. as Fig.17)..18.2 2.0 0.8 -0. for the optimum numerical treatment.2 0.2 0.6 0. All the previous computations were performed on a uniform panel distribution.9%.Fig.6 -0. Although the plots do not emphasize major differences between the two numerical solutions.005 -0. 15.13 presents a comparison between the wave cut profiles plotted on the hull based on computations over a hull paneled with 30…55 stations. A last set of computations to which a hull and free-surface nonuniform discretization is further proposed. Comparison between the wave cuts on the hull computed for hull discretization with 30.010 0..010 0..6 0.15) compared to cases 4 (Fig. for which the wave profiles were plotted.6 Figure 14.. establishing the discretization parameters (panel number. they vanish almost completely at the downstream.0 2. they exist.4 1. for a number of 45 panels. Because of that. 0.8 x/L 1.4 0. 8 and 12. going down to 1. It is worthy to point out that when the hull panel number decreases. Applying it to the ship speeds of 14. a complete set of computations were performed for a speed range between 14 and 18 knots. were reported to the standard case computation with 48 panels on the hull (case 8 in Table 1).55 de stations 0. which presents the variation of the wave coefficient with the panel resolution. This can also be seen in Fig.2 1. with a modification step of 0. Even though the free-surface profiles display several differences on the hull.6 1.14 bears out.4 0.8 2. Comparison between the free surface topologies computed for hull discretization with 30 and 48 stations .4 x/L 2.4 -0. panel minimum size).010 0. see Table 6.015 h/L 30 stations 35 stations 40 stations 45 stations 50 stations 55 stations 0.16) and 12 (Fig. Fig. as the Fig.005 0.13 shows.. having the computational domain size previously investigated.005 0.2 0.015 h/L 30 stations 35 stations 40 stations 45 stations 50 stations 55 stations 0. the final conclusions of the present study will be drawn for the computed solution over a hull discretized with 48 panels.010 -0.17 show the free-surface topologies computed for the nominal speed.8 1. Based on the previous discussions.

0 x/L Figure 18. Fig. The computations performed on the clustered panel distribution were denoted by . the difference between the two sets of solutions is significant.15. A great deal of numerical simulations was performed to clarify the influences of the numerical and geometrical parameters on the overall accuracy of the Shipflow solver.005 -0.Figure 15. The numerical solutions are presented in the following figures.20 shows a comparison between the free-surface topologies computed in the cases 33.000 -0.19 comparatively presents the two panel distributions.010 0.005 0.21 prove.0 1. Figure 16. the speed step being again 0. Comparison between the uniform and non-uniform panel distributions in the free-surface This fact will determine to a higher wave resistance.015 v= 14 Nd v= 16 Nd v= 18 Nd 0. Fig.0 2.010 0. and 33…41. 4. 16 and 17. Figure 17.5 knot. Thus. The computational method was based on a hybrid BEM-RANSE method that employs the zonal approach. respectively. as the Table 7 and the Fig. Comparison between the free-surface spectra computed in cases 4 and 8 Figure 19. Free-surface spectra computed in case 8 33…41 in Table 1 and they correspond to the same speed range of 14…18 knots. ACKNOWLEDGMENTS The reported research could not be possible without the generosity of the Romanian National University Research Council (CNCSIS) which is greatly acknowledged for the provided financial support through the Grant A_C code 33/2005. Comparison between the free-surface spectra computed in cases 8 and 12 h/L 0. Comparing them to those presented in Fig. one may see that the waves are higher in magnitude. a fact that may only be clarified after the experimental validation of the numerical simulation. 37 and 41. Comparison between the wave profiles computed in cases 8 and 12 Fig.21 depicts a comparison of the wave-resistance graphs determined in the computational cases 4…12. As it may be seen in the figure. CONCLUSIONS The paper presented a methodology for computing the 3D incompressible free-surface flow around a 10000 DwT LPG ship.

9697 Cw No Panel Clustering With Panel Clustering 0. AIAA-2003-0844.0 Figure 21. Comparison between the free-surface topologies computed on uniform and clustered panel distributions on the free-surface Table 7.5 0.R.75364 17 0.5 15 15. NM.72575 Cw×103 0.5 16 0.0020 16. (2003). 3. P. 2.0 15. pp. The wave resistance variation with the ship speed. W..79931 17.Figure 20.0 14. AIAA Committee on Standards for CFD-Status and Plans. Fluids Eng.0010 0.5 15. P. Perspective: A Method for Uniform Reporting of Grid Refinement Studies. R. Verification and Validation in Computational Science and Engineering.5 18. Computations on uniform and clustered panel distributions on the free-surface REFERENCES 1.5 1. (1994).5 17.0 16. C. Hermosa publishers.65092 0. Oberkampf. NV .0005 v [Knot] 14. The wave resistance variation with the ship speed Speed 14 14.P. 116. Roache. Cosner.5564 0.0 17.0015 0. 41st Aerospace Sciences Meeting. ASME J.F. 405-413. Vol.30101 18 1.. Rahaim.5 16. J.70339 0. (1998). Reno.. D. Albuquerque. Roache. Dominik..60721 0.L.

3. Wilson. Stern. Roy.J. NY. Casey. Wilson. RANS CFD Prediction of Pitch and Heave Ship Motions in Head Seas.G. AIAA-G-077-1998. Wilson. Wiley.W. F. Wilson... C. 1. 313-325. 15. F. Experimentation and Uncertainty Analysis for Engineers. E. No.. F. Roy. Aerospace Industry..and Two-Equation Turbulence Models for Hypersonic Transitional Flows.. R. J.. Flow Structure Around an Appended Tanker Hull Form in Simple Maneuvering Conditions.. Assessment of One. Benchmarking of Computational Fluid Dynamics for Ship Flows: The Gothenburg 2000 Workshop. 795-803. E. ERCOFTAC Special Interest Group on Quality and Trust in Industrial CFD – Best Practice Guidelines. V. No. F.803-810. Bertram.. Version 1. published by ERCOFTAC. New York. Busan. 32. F.V.. and Propeller on Forces and Moments. 63-81. 119. Raad.. 17. 7. (1997). U. Freitas. 123. M.. Nevada. Korea.4. Stern. W. Vol. F. Issue 4. Coleman. AIAA 2003-410 Applied Aerodynamics Special Session on CFD Uncertainty. (2003). (2003). (2003). Fluids Eng. (2004). E.. Larsson. Paterson. C. Coleman. F. Coleman. Fluids Eng. H. F. 11. P. Reno. Stern. (2003). Stern.. pp. Wilson. (2003).. Rudder Angle. H. T. Stern.). Reno. F. 10. C. R. Ghia. (1999). Steele.. Vol.. Vol. 5. 2nd Edition. 19. Weymouth G. Stern. R. Proceedings CHT-04... (2003). NV. R. (2003). Division for Fluid Dynamics.V. Uncertainties in CFD Code Validation. Korea. 16. ASME’s Quest to Quantify Numerical Uncertainty. Paterson. Paterson. Stern. 9. Verification of Codes and Solutions in Computational Simulation.. 20. Statistical Approach to CFD Code Certification (Invited Paper). Comprehensive Approach to Verification and Validation of CFD Simulations-Part 1: Methodology and Procedures. Busan.. pp. Computers and Fluids. J. AIAA 2002-0904 Aerospace Sciences Meeting. Wilson. Xing. pp. Wintergeste. pp. pp. Vol. H. (2003). 8th International Conference on Numerical Ship Hydrodynamics. Sub-Visual and Acoustic Modeling for Ducted Marine Propulsor. C. Roache.W. Simonsen. Busan. 41st Aerospace Sciences Meeting.. Fluids Eng. 13. F.. (2002). and Stern.. pp.... R.G. AIAA (1998). Stern. Stern. ASME J. (2000). (2001). C. Guide for the Verification and Validation of Computational Fluid Dynamics Simulations. Comprehensive Approach to Verification and Validation of CFD Simulations-Part 2: Application for RANS Simulation of a Cargo/Container Ship. Verification and Validation for RANS Simulation of a Naval Surface Combatant. M.. Standards for CFD in the 14. 13251356. 42nd AIAA Aerospace Sciences Meeting.. Shao. DES and RANS of Unsteady Free-surface Flows. Kim.. Celik. H. Vol. 12. (eds.. ASME J. Reno. Reno.. Simonsen. Journal of Spacecraft and Rockets. 123. 6. Journal Ship Research. Kandasamy. Stern. ASME J.. 8th International Conference on Numerical Ship Hydrodynamics. L. F. NV .. Coleman. 18. Korea. P.J. 8th International Conference on Numerical Ship Hydrodynamics. (2001). R. 40. 47.. T. I. (2004). Norway. F... 41st Aerospace Sciences Meeting. Verification and Validation of RANS Maneuvering Simulation of Esso Osaka: Effects of Drift.. NV 8..793-802. Blottner..

2007 NUMERICAL FLOW INVESTIGATIONS OVER A CHEMICAL TANKER HULL Ana-Maria TOCU. The computations were performed with the SHIPFLOW code that uses the zonal approach concept.tocu@ugal. viscous turbulent flow. U i Fluctuating velocity components Cartesian directions P Instantaneous pressure Time average pressure.Student* Department of Ship Hydrodynamics “Dunarea de Jos” University of Galati Adrian LUNGU. energy k Turbulent kinetic energy Specific dissipation of turbulent kinetic Dissipation of turbulent kinetic energy Normal to surface Parameter direction crossing the boundary Frictional resistance coefficient Viscous pressure resistance coefficient Viscous resistance coefficient Wave resistance coefficient Total resistance coefficient Form factor ε ni ξB CF CPV CV CW CT K φ φ t General variable Time average of φ Time . In CHAPMAN. Turbulence is treated with the k-ω SST model without wall functions.: (+40) 236 495400. Email: ana. In the zonal approach. Professor Department of Ship Hydrodynamics “Dunarea de Jos” University of Galati *Corresponding author: 47 Domneasca Street. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. KEYWORDS Potential flow. boundary layer. Flux-correction is used to increase the accuracy to second order in regions of smooth flow. µ ρ Kronecker’s delta µT ν δ ij Reynolds stress tensor Strain-rate S ij ω free-surface. Ph. the flow over the fore part of the hull is computed with a coupled potential flow-boundary layer method while the flow over the aft part with a RANS method. The free surface geometry is obtained as the potential-flow solution and it is kept fixed for the solution of the RANS equations. while the second-order flux corrections are used as an explicit defect correction. Romania Tel. Fax: (+40) 236 495400. The tri-diagonal system contains the first-order Roe convective terms and the second order diffusive terms. central differences around the cell face centers are used. the convective fluxes are evaluated using the approximate Riemann solver of Roe discretization. The discrete equations are solved iteratively with ADI solver.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). For the evaluation of the diffusive fluxes. In the present work the RANS solver CHAPMAN is used. 800008.ro ABSTRACT A numerical method has been employed for the prediction of the flow past a chemical tanker hull. Fascicola 3. RANSE. Romania June 1 .Galati.2. NOMENCLATURE LPP Length between perpendiculars B Moulded breadth Mean Draught T Instantaneous velocity components in Cartesian directions ui Time average velocity components in Cartesian directions. P p xi Ui Cartesian coordinates u i" in p" Ri σ ij ρ µ Fluctuating pressure Volume force Total stress tensor Density Dynamic viscosity Turbulent dynamic viscosity Kinematic viscosity.D.

B=18m. D=8.1. and that they communicate with the other components in the chain. For the aft part of the ship. Therefore. A flux correction with a min-mod limiter is used for increasing the accuracy to second order in regions of smooth flow. the viscous flow was predicted using the RANS code with boundary conditions defined by the potential-flow treatment as well as by the boundary layer computation. In CHAPMAN. and maintenance costs. performance safety and environmental elements. Speed= 12knot. In the present work. The initial design stage is important for a successful design. 2. pressure and turbulence (with a k-ω SST turbulence model) are fully coupled and solved. the potential flow codes have been used in order to provide immediate results. operation. but also the boundary layer parameters on the fore half of the ship. T=6. the equations for the momentum. Solving the RANS equations will give the time average velocity and pressure. The computations were performed for the ship having the main particulars: LPP=109.1. For hydrodynamic assessment of ship hulls in the optimization process. Knowing the average will usually be enough because of the time fluctuating velocity and pressure. INTRODUCTION The design of the ship affects directly the quality. By using information from a hull database it is possible to perform parametric hull optimization at a very early stage of the design process. testing the sensitivity brought by the choice of panel distribution and panel type. these required a greater effort to achieve solutions that lend them to practical application in an optimization environment. are much smaller in amplitude.1. that is why the changes in density are negligible. First. The continuity equation states that mass is conserved: 1 ∂p ∂U i + =0 ρ ∂t ∂xi (1) Incompressible flow is considered. It has a significant impact on production. SHIPFLOW includes two RANS codes. CFD codes play an important role. The continuity equation can be written: ∂U i =0 ∂xi (2) The Navier-Stokes equations of motion can be written in the following form: Fig. fast. some numerical solutions of flow around a realistic hull form are described. Their use in the optimization chain requires that they are accurate. reliable. RANS equations Even if the Navier-Stokes equations can be solved numerically by resolving all scales. Lines plan ρ ∂ U jU i ∂σ ij ∂U i +ρ = ρRi + ∂t ∂x j ∂x j ( ) (3) . In the first stage. in general. Due to the substantially higher computational effort associated with RANSE computations. the tanker has been used for testing the boundary layer and the Navier-Stokes viscous solvers. the tanker has been the subject of the calculations carried out based on the potential flow theory. which. Roe discretization is employed for the convective fluxes and central differencing around the cell face center is used for the diffusive fluxes. wave resistance. turbulent flows require extremely dense grids for the smallest turbulent length scales to be resolved. Then. The potential-flow theory has been used to compute waves patterns.8m. and the one used in the present work is CHAPMAN. we have to rely on the Reynolds decomposition in the foreseeable future.3m. The lines plan is depicted in Fig. The main attention was given to the stern where the viscous effects are significant.3m. which predicts not only the transition. NUMERICAL METHOD 2.1. It is essential to start with as optimal parameters as possible. lift and induced resistance and to provide the input to the boundary layer method.

ui" : The time averaged continuity equation and NavierStokes equations for incompressible flow can be written: ∂u i =0 ∂xi ⎛ ⎞⎞ ∂u (u u + u u ) 1 ∂p ⎛ ⎜υ ⎜ ∂u + ∂u ⎟ ⎟ + =R − i j i " " j i (14) U i = U i + u i" = u i + u i" (7) ∂t The instant pressure. where ε is the rate at which turbulent energy is dissipated by the action of viscosity on the smallest eddies. unrealistic results are obtained. φ1 = φ1 . ∂φ1 ∂φ1 . = ∂s ∂s φ1φ 2 = φ1φ 2 . in time mean velocity. [1].2. and time fluctuating velocity. Ui. departs significantly from its . where ω is a frequency of the large eddies [2]. The k-ω model performs very well close to walls in boundary layer flows. This model has been shown to eliminate the free stream sensitivity problem without sacrificing the k-ω near wall performance. p”: ∂x j i ⎜ ρ ∂x j ⎜ ⎝ ⎝ ∂x j i ⎟ ∂xi ⎟ ⎠⎠ j where P = P + p" = p + p" The time mean of a variable is defined as: (8) υ= µ ρ (15) φ = lim 1 T →∞ 2T T −T ∫ φdt (9) The following rules of averaging apply for any two turbulent quantities φ1 and φ 2 : φ1 = φ1 + φ1" . components are the mean pressure. The k-ε model is less sensitive to free stream values but generally inadequate in adverse pressure gradients and so Menter [3] has proposed a model which retains the properties of k-ω close to the wall and gradually blends into the k-ε model away from the wall. Turbulence models The most popular versions of two equation models are the k-ε model. This occurs when the ratio of the production of turbulence energy to the rate at which it is dissipated at the small scales. particularly under strong adverse pressure gradients. taking the time average of the Navier-Stokes equations (first moving all terms to the left hand side): ρ ∂ U jU i ∂U i ∂P ∂ +ρ + ρRi + −− ∂t ∂x j ∂xi ∂x j σ ij = − Pδ ij + 2µ ⎜ S ij − S kk δ ij ⎟ Sij is the strain-rate defined as: ⎛ ⎝ 1 3 ⎞ ⎠ (4) ( ) S ij = ∂U j 1⎛ ⎜ ∂U i + 2⎜ ∂xi ⎝ ∂x j ⎞ ⎟ ⎟ ⎠ =ρ ∂ U jU i ∂U i ∂P ∂ +ρ − ρ Ri + − ∂t ∂x j ∂x j ∂x j ( ) ⎛ ⎛ ∂U ∂U j ⎜ µ⎜ i + ⎜ ⎜ ∂x j ∂xi ⎝ ⎝ ⎞⎞ ⎟⎟ = ⎟⎟ ⎠⎠ (5) =ρ ⎛ ⎛ ∂U ∂U j ⎞ ⎞ ⎜ µ⎜ i + ⎟⎟ = ⎟ ⎜ ⎜ ∂x j ∂xi ⎟ ⎠⎠ ⎝ ⎝ ⎛ ⎛ ∂u ∂u j ⎞ ⎞ ⎜ µ⎜ i + ⎟⎟ ⎜ ⎜ ∂x j ∂xi ⎟ ⎟ ⎠⎠ ⎝ ⎝ (13) Skk in (4) is zero for incompressible flow ⎛ ⎟ ⎜ u j ui + u "j ui" ⎞ ∂ui ⎠ − ρ R + ∂p − ∂ +ρ⎝ i ∂x j ∂xi ∂x j ∂t S kk = 1 ⎛ ∂U ki ∂U k ⎜ + 2⎜ ∂x k ⎝ ∂x k ⎞ ∂U k ⎟ ⎟ = ∂x = 0 k ⎠ (6) The Reynolds averaged Navier-Stokes equations can be derived from (3) by splitting the instant velocity components. " φ1φ 2 = φ1φ 2 + φ1"φ 2 . However it is very sensitive to the free stream value of ω and unless great care is taken in setting this value. φ1" = 0 . P.σ ij is the total stress and for a Newtonian fluid can be written: Then. The performance of two-equation turbulence models deteriorates when the turbulence structure is no longer close to local equilibrium. p. and time fluctuating pressure. and the k-ω model. Taking the time average of the continuity equation gives: ∂U i ∂U i ∂u i = = =0 ∂xi ∂xi ∂xi (11) Subtracting (11) from (2) gives that also the time fluctuating velocity fulfills the incompressible continuity equation: ∂u i" =0 ∂xi (12) 2. (10) φ1 + φ 2 = φ1 + φ 2 . ui. φ1" φ 2 = 0 .

.9346·10-3 0.9375·10-3 3805 4285 4765 The potential-flow method is used to solve the free surface problem and to provide the input to the boundary layer method. at the velocity v=11 knot. GRID GENERATION A 3D grid is generated by using suitable parameters for the RANSE solver. Computed wave coefficient Total number of Wave resistance panels coefficient Cw -2 0. The turbulent kinetic energy is over-predicted in regions of flow impingement and reattachment leading to poor prediction of the development of flow around leading edges and bluff bodies. For instance. 4. It is noted that the grid covers only the aft half on one side of the hull (see Fig. The grid generator will automatically choose a suitable height. Because the Chapman solver does not use wall laws. the program computes the outer boundaries. The free-surface problem is nonlinear since the free-surface boundary conditions are nonlinear and must be satisfied on the initially unknown wavy free surface. Table 1. In each iteration the wave height is computed from the linearized dynamic free surface boundary condition.2). outside of this region. ω = f (uτ . The potential solving module has been used on different Reynolds numbers and in different calculation grids cases. while the dynamic pressure has the zero value. The offset file has been prepared in order to discretize the hull for the numerical simulation.e.1. The present analysis describes the flow around a three dimensional body. leads to marked improvements in performance for non-equilibrium boundary layer regions such as may be found close to separation.1001·10-2 3325 v = 11 knot 0. In addition. Therefore. At the upstream the oncoming flow velocity is supposed constant. the k−ω−SST model is used having k−ω model near the wall and a k−ε model. Boundary conditions The boundary conditions for computing the flow within the solution domain are requiring the no-slip condition on the hull surface for the velocity. . The solution method for the non-linear problem is used to linearize the free-surface boundary condition around a known base solution and to solve the problem in an iterative manner. absolute value of the rate of strain times k/ε) become large.9406·10-3 0. Various attempts have been made to modify two equation turbulence models to account for strong nonequilibrium effects.) . The potential flow and the boundary layer solution are utilized for computing the initial flow profiles. 2. The turbulent quantities are computed from analytical formulas based on the velocity profile in the inlet plane. transformed to resemble a k−ω model. as k and ω are. whereas the pressure is extrapolated with zerogradient. Kato and Launder [4] have proposed a modification to the transport equation for ε which is designed to tackle this problem. The SST version of Menter k-ω based also offers a considerable improvement. the velocity k and ω are extrapolated with zero-gradient. a Neumann condition for the pressure. In order to generate this single block grid. COMPUTATIONAL STRATEGY An offset file based on the initial lines plan. the results show that the level of grid dependency is large. with free surface. at multiple ship speeds. For example. In the symmetry plane. as shown in Fig. while for k and ω the following Dirichlet conditions are imposed: k = 0 . the momentum integral equations for the boundary layer are solved along streamlines traced from a potential-flow computation. defines the geometry of the ship. 3.1596·10 2920 0. and with sinkage and trim.. This step is used to obtain a smooth distribution of the turbulent quantities. as Table 1 proves. or equivalently when dimensionless strain rates (i. the so-called SST (shear stress transport) variation of the Menter model. with transom stern. For thin turbulent boundary layer computations. One single structured grid having the inner surface fitted to the hull will be created. the height of the cells closest to the hull surface should be very thin.. velocity components are extracted from the boundary layer computation and outside the boundary layer from the potential flow solution.3. At the downstream.“equilibrium value”. always controlling the functions to achieve orthogonality. the equations for the turbulent quantities are solved assuming a zero velocity gradient in the main flow direction. for different panels number. in the present work. For the viscous calculations. zerogradient Neumann conditions are imposed for all the variables. Steady state incompressible flow in a coordinate system that moves with the body is assumed. [3].

an introspection within those parameters fields is proposed.2 Three dimensional grid topology generated around the aft part of the hull 5. as expected. NUMERICAL SOLUTIONS One of the most important issue of each numerical simulation output is the detailed analysis of the physical parameters that describe the flow. The boundary layer module predicts transition and boundary layer parameters on the forward half of the ship.4 Free surface topology Fig. respectively. For being more precise. Some other results from the potential flow module.. This behavior is confirmed by the physics behind the phenomenon. non-linear computations were employed for finding out the distribution of the pressure and velocity components on the body. . causes flow suction. 5. Therefore. The increase of the speed determines an increase of the wave length. Verifications confirmed that the computed solution agree with the theory. In Fig. Fig.3. under the bulb stern. the flow becomes unstable downstream the hull. i. computed for the 12 knots velocity case are depicted in Fig.Fig.4 and Fig. the dynamic pressure coefficient computed for 12 knots is shown. Various simulations were performed to check the performances of the lines plan. λ=2π•L•Fn2. The negative pressure field observed at the aft part in the region of the bottom.6.3 Dynamic pressure coefficient on the hull.7 the values for the boundary layer thickness are depicted. The resulted pressure field shown in Fig. can be later on used in the hull form optimization process. including the free surface elevation and the wave profile. which cumulates to that created by the propeller. where the transom determines the generation of wave crests of a significant magnitude.5 Wave profile non-linearly computed for a speed of 12 knot It is worth mentioning that when the speed exceeds 13 Knots. In Fig. Fig.e.

6 Pressure distribution on the hull structure and wave propagation (i. a satisfactory position has been achieved. . CONCLUSIONS For the development of new hull forms it has become increasingly important to predict numerically the hydrodynamic properties of a new vessel at an early stage of the design process.568·10-3 CPV 0. Realistic estimates of total resistance can be achieved using a combination of a non-linear potential flow solution for the free surface.979·10-3 K 0. the RANS solver provides pressure and velocity distribution on the hull. but for certain vessels. for the optimization of hull forms.243·10-3 CW 0. Viscous calculations. Reliability and accuracy of CFD calculations are essential prerequisites for the improvement and. On the basis of numerical flow field analysis. The potential flow discussed earlier.8).395 6.Fig. Table 2 CF 1. Prediction of stern wave patterns. as may be seen in Fig.10.9 Velocity vectors and pressure distribution on the aft body.9. which once converged that can be used as a fixed free slip boundary for the RANSE calculation. The grid used for viscous computations has 120x30x60 grid nodes in the longitudinal. without numerical diffusion or dispersion). is still a subject of research. particularly with respect to transom sterns. the most promising design alternatives must be chosen before model tests are conducted.6748·10-3 CV 2. respectively (see Fig. The iso-lines of the axial velocity component for some arbitrary slices are also depicted in Fig. has demonstrated accuracy in both wave pattern Fig. The validation status on non-linear free-surface potential flow codes for wave resistance is considered to be in a relatively satisfactory state. Resistance results obtained from the viscous calculations are shown in Table 2. Fig.e. transverse and normal direction.8 Pressure coefficient in the aft hull obtained with the RANSE solver Fig. finally.7363·10-3 CT 2.7 Boundary layer thickness Through an iterative process.

. Launder.11.1974 2. Regnstrom. 9th Symposium on Turbulent Shear Flows. Vol.3 . Menter.4. Patel. G. Tzabiras. Dick.. D.. J.4. G. Patel. (1985). Faculty of Naval Architecture. B. E. Deng..1-10..Users manual 12. (1988) Ship Stern and Wake Flows: Solutions of the Fully-Elliptic Reynolds-Averaged NavierStokes Equations and Comparisons with Experiments.. J. Iowa Institute of Hydraulic Research Report No. Visonneau. Zonal Two Equation k-ω Turbulence Models for Aerodynamic Flows. J.14. Kato. Three-Dimensional Flow Computation with Reynolds Stress and Algebraic Stress Models 7.B. Flowtech International AB (2006) SHIPFLOW 3. (1974) The Numerical Computation of Turbulent Flows. (1998) Some Improvements in Menter’s k-ω SST Turbulence Model. 1311-1323 8. Merci.. D. Dick. (2005) Viscous-Flow Computations of Two Existing Vessels at Model and Full Scale Ship Reynolds Numbers. In 29th Fluid Dynamics Conference. . S. 10. Orlando. REFERENCES 1. International Journal for Numerical Methods in Fluids. 2085-2093 10. M.. AIAA paper-98-2554 9..R. Spalding.. AIAA Journal. Ju. AIAA paper-93-2906 4. 10 Iso-wakes at different planes ACKNOWLEDGMENTS The reported research could not be possible without the generosity of the Romanian National University Research Council (CNCSIS) which is greatly acknowledged for the provided financial support through the Grant A_C code 33/2005. (1985) A New Class of High Accuracy TVD Schemes for Hyperbolic Conservation Laws. A. pp. . Wilcox. International Conference on Computational Methods in Marine Engineering. C. B. Computer Methods in Applied Mechanics. pp. (1985) Computational Treatment of Source Terms in Two-Equation Turbulence models.. S.. B..Fig. No. Hellsten. vol. Launder. E. Chen. Proc. California 3. Kyoto. All the numerical computations have been performed at the Department of Hydrodynamics from the “Dunarea de Jos” University of Galati. S. B. Albuquerque. Vol.. AIAA paper-85-0363 6.38..B. V. Barcelona. W. V. G. (1993)The Modeling of Turbulent Flow Around Stationary and Vibrating Square Cylinders.C.C.E. 323 13.. Starke. .R. E. H. 3. A.. Sheuerer. Reimslaugh. M.. Turbulence Models for Near Wall and Low-Reynolds Number Flows: a ... F. (1993) Turbulence Modeling for CFD.. In 24th Fluid Dynamics Conference. (1992) A Multigrid Method for Steady Incompressible NavierStokes Equations Based on Flux Difference Splitting. Queutey.6 5. Schweighofer. Linden. NM.C.R. Chakravarthy.. Vierendeels... RODI. Osher. 269-289. P. K.. 2005 11..

theoretical manual. and Stern. M. 1. Larsson.. 63-81.14. R.. n. AIAA Journal.. 5-8 Jan 2004.23. 41st Aerospace Sciences Meeting.. 17. Stern. pp. Nevada. pp.. Stern.. (2003). L. F. Wilson.1308-1319 15. Reno. No. (2000). F. Nevada. 6-9 January 2003 . and Bertram.. Kandasamy. F.. March 2003. V.. Benchmarking of Computational Fluid Dynamics for Ship Flows: The Gothenburg 2000 Workshop. T. J. Vol. 47. XCHAP . and Shao. Review.9. R. Reno. Xing. 18. 42nd AIAA Aerospace Sciences Meeting. 2005 16. Division for Fluid Dynamics. Journal Ship Research. Flowtech International AB (2006). (2004) DES and RANS of Unsteady Free-surface Flows. AIAA 2003410 Applied Aerodynamics Special Session on CFD Uncertainty. vol. Statistical Approach to CFD Code Certification (Invited Paper). Wilson.

The tendency toward increasing the ships speeds demands new types of propulsion devices which operate satisfactorily at high speeds. a discussion of applications of propellers and waterjet systems for high speed ships in light of advantages and disadvantages of these propulsion devices is given. Alternatively. the interest for high speed ships grew significantly worldwide. cavitation. The propulsion device must be selected by carrying out a study that balances the ABSTRACT The paper is focused on performances evaluation of propulsion systems for high speed ships. Romania July 1-2. Hybrid propulsions systems of waterjet and conventional fixed. waterjet propulsion systems are serious competitors to propellers. Propellers design for high speed ships meets difficulties mainly due to presence of cavitation on propeller blades.65 Q1 / 2 H 3/ 4 Subscripts and Superscripts . INTRODUCTION During the last decades. propeller cavitation cannot be avoided and the problem is to select and design a propeller able to accommodate the detrimental effects of cavitations: loss of thrust. [1. high speed propellers can be categorized in cavitating and supercavitating according to the development of cavitation. with higher efficiency. NOMENCLATURE D [m] diameter revolution rate n [s-1] V [knots] ship velocity [m/s] advance velocity Va [-] propeller blade area ratio Ae/A0 P/D [-] propeller pitch ratio z [-] number of blades [N/m2] total static pressure p0 [N/m2] vapour pressure of water pv q [N/m2] dynamic pressure Q [m3/s] volume flow rate [kW] delivery power PD η [-] propulsor efficiency σ= p0 − pv [-] q V [-] J= A nD cavitation number propeller advance coefficient specified speed n s = 3. supercavitating propeller. It is found to be more expensive and more complicated than a propeller. Secondary.: (+40) 236 495400. As an alternative for propeller cavitation and vibrations problems. waterjet propulsion systems are now applied to a wide range of ships types. However. controllable pitch propellers or contrarotating propellers combine the bests of both.ro P propeller R waterjet rotor 1. The propulsion device must be selected very early in the high ship design process and the type of propulsor can have a strong impact on the ship design itself [3]. military and superyacht market. For high speeds ranging up 20-25 knots. erosions. Fax: (+40) 236 495400. The propellers are used for normal cruising speeds. Galati. the results for cavitating propellers and axial jet propulsion systems are presented. while the combination of waterjet with the propellers is used for maximum speed requirements. an application related to propulsion performances for a ship operating in the speed range of 30-38 knots is discussed. 2006 PERFORMANCES DATA OF PROPULSION SYSTEMS FOR HIGH SPEED SHIPS Mihaela AMORARITEI. Romania Tel. surface-piercing propeller. noise and vibrations. Lecturer*. Naval Hydrodynamic Department “Dunarea de Jos ” University of Galati *Corresponding author: 47 Domnesca Street. Email: mamor@ugal.Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics Special issue 3rd Workshop on Vortex Dominated Flows Bucharest. low noise and vibrations.2]. First. In these cases. KEYWORDS Propeller. A condensed survey on high speed propellers problems is presented. 800008. with a speed range above 45 knots. Fast ships can offer important advantages in commercial. waterjet.

The leading edge vortex cavitation may have an appearance which is much like sheet cavitation. Any great extent of cavitation affects the flow around propeller blades. Hub vortex cavitation occurs in the vortices shed in the vicinity of the blades roots. erosion. the torque and the efficiency of the propeller are reduced. other types cause pitting of the screw metal and accelerate blade erosion. Alternatively. The propeller hull vortex cavitation occurs when a strong wake peak interacts with the propeller and it causes irregular vibrations and an extremely high noise level. The end of tip vortex cavitation is far downstream and gives an excellence visual demonstration of the contraction of the race column behind the propeller. According to the nature of cavities and to the particular location on the propeller. Sheet cavitation is characterized by a continuous liquid/vapor interface attached to the blade surface. The paper presents aspects of performances evaluation of propulsion systems for high speed ships. noise and vibrations. 2. First. suppression of cavitation becomes impossible. A condensed survey on high speed propellers problems is presented. . Cavitation . local tip vortex cavitation. A sketch of different types of cavitation on a marine propeller is presented in Figure 1 (taken from [5]). Cavitation types. a discussion of applications of propellers and waterjet systems for high speed ships in light of advantages and disadvantages of these propulsion devices is given. noise and vibrations are the major problems of propellers for high speed ships. . erosions. Secondary. These types of cavitations have different appearances and undesirable effects: some types of cavitation have major contributions to hull vibrations or high frequency noises. cavitation and vibration characteristics. Figure 1. Propeller hull vortex cavitation is a special form of vortex cavitation which extends from the propeller to the hull of the ship. the results for cavitating propellers and axial jet propulsion systems are presented. low noise and vibrations level.cloud cavitation . .the phenomenon that water changes its phases into vapor in flow regions with very low pressures . ussualy at high leading edge loading conditions. the rotational velocity of propeller. costs. the thrust. These problems are caused by a “combination” of the operating conditions: the high speed of the ship. the large delivered power to the propeller and the non-uniform wake-fields in which the propeller operates behind the ship. A fully developed tip vortex cavitation is a major contributor to hull excitations. propeller-hull vortex cavitation. The trailing vortex cavitation incepts downstream of the blade tip and it is determined by the radial loading distribution. an application related to propulsion performances for a ship operating in the speed range of 30-38 knots is discussed. “Criteria must be given on the type of cavitation to avoid and amount of thrust breakdown to be allowed” [4]. Different types of cavitation are mentioned and the detrimental effects on marine propeller performances are discussed: loss of thrust. causes a change in the lift and drag of the propeller blade sections and as a result.root cavitation. The leading edge vortex cavitation is determined by the vorticity produced at the leading edge. local and leading edge vortex cavitation is presented in [6]. leading edge vortex cavitation. In what extent the vortex cavitation is important for rudder erosion and for pressure fluctuation on the hull represents an important subject of concern [6].vortex cavitation: trailing edge vortex cavitation. Generally. with satisfactory efficiency.has detrimental effects on propeller characteristics: loss of thrust and torque. hub vortex cavitation.bubble cavitation. A detailed discussion regarding distinction between trailing. Sheet cavitation usually begins at the leading edge . [5] Vortex cavitation requires the presence of vorticity.propulsive efficiency. and it is necessary to select and design several screw type propellers able to operate at low cavitation numbers.sheet cavitation (face / back cavitation). noise and vibrations. different types of cavitation occur on marine propellers: . the hub vortex does not contribute to hull vibration unless it is very strong and excites the rudder [7]. HIGH SPEED PROPELLERS Cavitation. at low speed and high loaded conditions. For high speed above 20-25 knots. The local tip vortex cavitation occurs in the vortices shed in the vicinity of the blade tip and it is influenced by the tip geometry. propeller submergence depth.

all designers of screw propellers for low and moderate speed ships endeavour to keep the propeller free of cavitation: the small amount of cavitation that does occur will not affect performances nor cause noise. the most important problem is to control the influence of cavitation rather than attempt to suppress its occurrence. A partial cavity is a cavity that is shorter than the chord length of the blade. If partial cavitation cannot be avoided it is preferable to use special propellers designed for . From the point of view of pitting of the blades and erosion the following types of cavitations have been identified as major contributors: bubble and cloud cavitation. In the design range 0. When the propeller operates in non-uniform wake behind ship. propeller hull vortex cavitation. Chart of practical use of supercavitating propeller [12] “Experimental results have indicated that in order to have satisfactory supercavitating operations. if the angle of attack is positive. These cavities grow larger. For high speed propellers.045.7 relative radius should been less than 0.045<σ<1. near the midchord or maximum thickness of the blade. In bubble cavitation. isolated spherical cavities are formed in the fluid at points where the pressure falls close the vapor pressure. tip vortex and collapsing sheet cavitation off the blade. the cavitation number of the blade section at 0. Figure 2. o supercavitating propeller: fully submerged supercavitating propeller surface . when “both the advance and rotational speed are so high and it is extremely difficult to avoid cavitation. The definition excludes the highly loaded merchant ship propeller [11]. The cloud or mist cavitation usually occurs behind the collapse of the sheet cavitation and causes blade erosion. sheet cavitation and unstable tip vortex cavitation. move downstream with the fluid and collapse as they enter a region of higher pressure. When the tip vortex cavitation and the sheet cavitation pass a strong wake peak. vibrations and erosion damage. It is known that sheet cavitation is further divided in partial cavitation and supercavitation. Some types of cavitation are mainly responsible for large forces exciting hull vibrations: suction side sheet cavitation on the blade. many of which operate under appreciable cavitating conditions [9]. sheet cavitations occurs on the propeller blade face. According to the development of cavitation and to the flow regime in which trey operate. The impact of collapsing bubble is very high and the bubble cavitation is considered the principal cause of erosion. they may break up into cloud and very small bubbles which collapse separately while moving with the fluid. this requires a high rotational speed” [12] . sheet cavitations occurs on the back of the blade.piercing propeller. A propeller belongs into the high speed category.(on either the face or the back side of the propeller blade). when the pressure distribution has an adverse pressure gradient. The effect of cavitation on hydrodynamic performances of a propeller is that increased power and rotational speed are required without a commensurable increase in the speed of the ship. The recommended field of applications for supercavitating propeller was given in Figure 2 taken from [12]. Generally. This happens when the angle of attack derivates from the ideal angle of attack: if the angle of attack is negative. The spread of sheet cavitation over the blade and the occurrence of bubble cavitation are accompanied by loss of thrust [8]. This periodic growth and collapse of sheet cavitation with the blade positions called intermittent or unsteady sheet cavitation was indicated as the cause of large forces exciting highly undesirable hull vibrations. and if the propeller operates at moderate speeds (35 to 50 knots). This does not apply for high speed propellers. the high speed propellers can be clasified in: o cavitating propellers: Newton Rader propeller. Gawn Burrill propeller. even under ideal inflow conditions” [10]. An important question in the initial design is whether a cavitationg or a supercavitating propeller is the most desirable for specified operating conditions of a high speed ship. any propeller designed to fully cavitation have a considerable lower efficiency than a cavitating propeller. A supercavity is a cavity that is longer that the chord length of the blade. The performance characteristics of high speed propellers depend to a considerable degree to the proper combination of the advance coefficient J and the cavitation number at which the propellers have to work. the sheet cavitation extent varies as the propeller rotates.

three and four blades propellers have been developed. In a supecavitating propeller.11) and in form of polynomials for mathematical description [13]. efficiency). The Figure 5. torque. Modified supercavitating blade sections : tripping wedge and tripping wire [17] To predict the supercavitating propellers performances in the initial design. the cavity on the back of blade has spread until it covers the whole of the back and it collapses far downstream to the trailing edge. extend and collapse downstream of the blade trailing edge.15]. The data from the systematic tests of model propeller series in cavitating environment are presented in graphical forms (Figure 10. Improved methods based on circulation theory.[16]. for extended periods at off design conditions.1. The Gawn Burrill series consists of three bladed propellers with elliptical developed blade outlines and segmental blade sections (Figure 3. The accepted effects of cavitation on Gawn Burrill and Newton Rader propellers are the changes in propeller hydrodynamic characteristics (thrust.48 to 0. significant cavitation and blade erosion cannot be avoided and it is then advantageous to consider the use of supecavitating propellers. The blade section shape used in Newton Rader series has the leading edge modified to achieve freedom from face cavitation (Figure 3b). The supercavitating propellers cannot offer advantages in noncavitating conditions and they are not suitable for running at lower speeds. Figure 4. . Estimation of the optimum diameterrotation speed relationship and selection of number of blades on the basis of stress and vibrations are two important problems in initial design of the supercavitating propellers. For very high speed ships (40 knots and higher). the advance coefficient and main propeller geometry (P/D. Blade sections for cavitating propeller (a) Gawn Burrill. polynomial equations for the series hydrodynamic characteristics have been obtained and the propeller of optimum efficiency can be automatically estimated using computed codes based on standard series.such conditions.08. Generally. At present. (b) Newton-Rader Newton Rader Series consist on three bladed propellers designed from a minimum cavitation number σ = 0. the second is to modify the back of the supercavitating section through a tripping wedge or a tripping wire so that separation of flow will occur (Figure 5). blade area ratios Ae/A0 from about 0. [13.25 and cover pitch ratios P/D from about 1. pitch ratios P/D = (0. series chart for two. Ae/A0). Figure 3.6 – 2) and blade area ratios Ae/A0 = (0. Using the method of the multiple linear regression analysis for each of series.5 to a value corresponding to atmospheric pressure.18) [13.04 to 2. supercavitating propellers are characterized by a very sharp leading edge and a very blunt trailing edge (Figure 4). The supercavitating propellers require appendages like shafts. The blade section shape in supercavitating propellers differs considerable from the conventional propeller. Gawn Burrill series were designed to cover cavitation number σ from about 0.51 .95. in subcavitating regime the efficiency is lower than for corresponding conventional propeller. non-linear cavity flow theory and boundary surface panel technique have been extended to evaluate the hydrodynamics performances of supercavitating propellers. One is to ventilate the cavity on the back of the propeller blade by introducing air. 14]. relatively a few cavitating propellers series have been developed and tested for a wide range of cavitation number. The cavitating propellers belonging to the Gawn Burrill and Newton Rader series are used for high speed crafts with a good efficiency during cavitating conditions. rudder and shaftstruts which lead to significant drag penalty. Some solutions have been proposed to avoid the lower efficiency under partially cavitating conditions and to extend the supercavitating propellers applications to lower speed. The thrust and torque coefficients are given as functions of cavitation number. Flow around blade section for supercavitating propeller Major advantages of supercavitating propellers in comparison with cavitating propellers include relatively high efficiency during fully supercavitating conditions and reduction in noise and blade surface erosion as a result of vapor cavities which originate at the leading edge of the blade.a). Other solution is the combination between supercavitating propellers for high speed operations conditions and subcavitating propellers for low speeds.

A special type of supercavitating propeller is the surface-piercing propeller also called partially submerged supercavitating propeller or ventilated propeller. there are cases when the local relative velocities along the propeller’s blade varies with radius and the cavitation number near the blade tip is to small for sub-cavitating conditions. In this case. The unsteady forces on propeller blades create relatively high vibrations and strength problems and a suitable distribution of skew may be defined to reduce the fluctuations of unsteady forces and reduce vibrations. A shaft angle of 10 degrees in high speed craft is commonly used for a sufficient aft draught. Surface piercing propeller blade section shape . The thrust and torque depend on propeller immersion. which takes water from outside in through a duct. The main obstacle for high speed propellers: cavitation is replaced by ventilation . struts) leading to reduction or complete avoidance of appendage resistance. lower ship resistance due to absence of underwater appendages like shafts. A surface-piercing propeller can have a large diameter. In this case. The absence of underwater appendages makes waterjet an ideal solution for shallow water operations. The model test confirmed the expected result that fluctuating forces are reduced in fully cavitating propellers. A trans-cavitating propeller (Figure 7 taken from [19]). yet that near the hub is to big for supercavitating conditions. its size is not limited by the blade tip clearance from the hull and by the vessel draught. The cyclic variations of angle of attack increases with the oblique flow angle and the advance coefficient. The propeller forces in oblique flow: (propeller thrust along the shaft. but are reduced remarkably with decreasing cavitation number”. 3. rudder and shaftstruts. the cavitation phenomena are intensified and contribute to propeller hull pressure augmentation. No reversing gearbox is required. The expected efficiency has been obtained in the design conditions. The fluctuating parts of the hydrodynamic forces and moments induced by propeller acting in oblique flow are very important with regard to noise and vibrations. adds energy and provides a jet reactive thrust of high velocity water expelled through a nozzle. The absence of underwater appendages (shafts. transverse force due to the oblique inflow) must be taken into account in prediction of the high speed ships performances. shaft angle and vessel’s trim. Partially submerged propellers have also problems with operations at low speeds and the artificially ventilation is a solution.diamond back [18] Even in high speed ships. . The disadvantages of surface-piercing propellers are the large blade impact forces when propeller blades enter water. the propeller shaft is just above the free surface and the supercavitating propellers operate at partially submerged conditions. ITTC [22] specified that “the thrust fluctuation for a single propeller blade on highspeed vessel can attain values of 80 to 100 percent of mean thrust of one blade. waterjet propulsion systems are now applied to a wide range of ships types and are becoming serious competitors to propellers. Taking into account the advantages and the disadvantages of supercavitating propellers. the excellent manoeuvrability at all ship speeds. the ability to accelerate. to place the propeller below the hull. reverse Figure 6. Trans-cavitating propeller blade [19] Most of high speed propellers are mounted on inclined shaft. a trans-cavitating propeller represents a new solution. torque. the thrust and torque fluctuations become larger.the phenomenon which refers to introduction of air into cavitation area due to the presence of the free surface. adopts non-cavitating sections near the root (domain A) and supercavitating blade sections near the tip (domain B). The waterjet unit is an impeller or pump inside the hull. Figure 7. the propeller operates in oblique flow which induces a cyclic variation of angle of attack and changes in propellers’ hydrodynamic characteristics. wedge shaped sections with cupped trailing edge similar to those used in fully cavitating conditions and a patented “diamond back” shape. Blade sections for surface piercing propellers include: wedge shaped sections. WATERJET PROPULSION As an alternative for propeller cavitation and vibrations problems. without any limitation on the size of pump. The advantages of surface piercing propellers in comparison with fully submerged propellers include the reduction of blade surface erosion since they are not susceptible to cavitation. The advantages are: higher efficiency.

with much lower noise and vibration levels. axial flow pump: 500< ns<1200. ship propulsion performances with cavitating propellers have been investigated. The waterjet propulsion unit occupies considerable space inside the ship and the impeller access for inspections and repair is poor. The components of a waterjet propulsion system are schematically presented by Kruppa in [10] (Figure 8). The excellent manoeuvrability is accomplished by directing the system's exit flow into any direction. The following range of ns specify the three types of pumps generally used: centrifugal pumps: 40< 34 240 2 38 285 2 First. the volum flow rate Q and the total head of pump H. The stator takes rotation out of the water and increase the pressure downstream the pump. ship which can be fitted with twin propellers or twin waterjets. mixed flow pump: 300< ns<500. High speed ships frequently use either of Gawn Burrill and Newton Rader propellers for which the hydrodynamics performances al low cavitation numbers are known. where high pump efficiency provides higher speeds with lower fuel consumption. both experimentally as well as numerically. centrifugal or mixed flow and these pumps have maximum efficiencies at certain range of specified speed ns as function of the rate of revolutions n. and the wake and thrust deduction fractions have been set to zero. Schematic waterjet propulsion system taken from [10] The inlet location depends on the hull and waterjet propulsion system configuration. APLICATIONS AND RESULTS The application is related to propulsion performances for a ship operating in the speed range of 30-38 knots. The jet reactive of high velocity water is expelled through a nozzle which converts the pressure into kinetic energy. Advantages of waterjets versus propellers are safety in vicinity of other crafts. Figure 9. The waterjet propulsion unit has higher weight than other propulsion devices: the weight of the water in the system above the free surface must be included in the system weight. ns<300. Regression polynomials for . Ship main characteristics Speed of ship [knots] 30 Resistance of ship [kN] 212 Number of propulsors 2 Figure 8. protection of marine life and environments and an expected reduction in acoustic signature. avoiding larger inlet losses and susceptibility of pump cavitation. Absorption of engine power is independent of ship speed and no hull vibration or high speed cavitation ensures comfort on board. Below 32 knots. The flow through the inlet duct can be characterized by the Inlet Velocity Ratio defined as ratio of the ship speed and the average axial velocity across the pump’s inlet. The main characteristics of ship are presented in Table 1. Table 1. There are some important disadvantages: it is found to be generally more expensive and the propulsor itself is more complicated that a conventional propeller.and crash-stop is much better as compared to propeller propulsion. Propulsion coefficient waterjet versus propeller taken from [24] 4. The waterjet propulsion system requires optimization from the inlet to the exhaust nozzle and a lot of research has been carried out to optimize the inlet and the duct shapes. Different types of pumps can be used: axial flow. Waterjets are ideal for speeds in excess of 35 knots. the advantages of waterjet versus propeller are a higher thrust and a rising propulsive coefficient PC (Figure 9)”[24]. the propeller has the advantage. “Above about 35 knots.

The results regarding evaluation of propulsion performances of cavitating propellers are presented in Table2.55 0.3 0.6 Newton Rader Gawn Burill Waterjet 0.18.77) 10KQ (0. The hydrodynamics characteristics of designed Gawn Burrill and Newton Rader propellers are presented in Figures 10. Propulsion coefficient waterjet versus propeller taken from [24] 5.02 nR [rpm] 780 740 Q 9. High speed ships frequently use Gawn Burrill and .7 0.4 0. The Gawn Burrill propeller suitable for 30 knots has a lower efficiency at 38 knots.61) 10KQ (0. P/D=1. Ae/Ao=0. Open water characteristics of Newton Rader propeller Table 3. P/D=1.77) 10KQ (0.77) KT (0. The propellers required to run above about 36 knots may be supercavitating.29 0.95 1.678 3488 Gawn Burrill P/D=1.10KQ 0.6 Newton Rader P/D=1.337 V 30 34 38 nP [rpm] 813 892 1000 0.648 η PD 2330 2851 36 1.337. Open water characteristics of Gawn Burrill propeller 0.09 700 13.279. Waterjet performances [26] Waterjet V 28 32 DR 0.615 0.3 0. It can bee seen that the propulsive efficiency of the selected waterjets exceeds that of the designed propellers about 34 knots and above. Ae/Ao=1.61) J 0.1 0 0.2 0.6 0. The results for axial jet propulsion systems are presented in table 3 [26].656 0. Ae/Ao=1.2 1.661 0.61) 10KQ (0.6 0.11.4 0.2 0.65 Figure 10. CONCLUSIONS Generally.4 J KT .613 η PD 2490 3185 4590 0. Ae/Ao=0.2. In this particular case.4 Figure 11.642 0.279 V 30 34 38 nP [rpm] 852 962 791 0.95 KT (1) 10KQ (1) KT (0.5 0.2 1. The diameters of rotors and the rotational speeds have been selected according with relation between power and ships speed for the most often used waterjets sizes. the advantages of propeller over waterjet are higher efficiency and low costs.22.7 0.5 0.6 0. Propellers performances Gawn-Burrill DP=1.10KQ 0.663 ηD PD 2546 3174 4199 KT .8 1 1.8 1 1.1 0 0.99 0.658 0.these series have been used to design optimum propellers at the speed design (34 knots) and to analyse their hydrodynamics performances in off design conditions.38 0. the propeller design for a high speed ship is a compromise of many considerations. η 0.18 KT (1) 10KQ (1) KT (0.61) The propulsion devices efficiency has been plotted in Figures 12.5 26 28 30 32 34 36 38 40 V [knots] Figure 12.42 11.77) KT (0.61 σ Newton – Rader DP=1. below 34 knots.77 0.95. Table 2.

The High Speed Surface Craft Conference. Kozhukharov. (1998). J. W. The Propulsion Committee Report 21. Y. (2001). R. The Design and Estimated Performance of a series of Supercavitating Propellers 13.Z.J. Aspects of Performance Evaluation of Waterjet Propulsion Systems and a Critical Review of State-of-the –Art. hydrofoils. Z.W. Haberman.D.P.B.(1967). CAV2001 7...B.. Netherlands. patrol boats.P. Cavitating Propellers Characteristics and their use in Propeller Design... 34th Wegemt School.A.Morgan... vol. A.No.P.G. low noise and vibrations. Design of Trans-Cavitating Propellers and Performance Analyses of the Test Result. Stefan Totolici (Bureau Veritas Romania) and to Mr. Ghose J.. Basic ship propulsion.K.W. The propulsion devices in service for high speed ships will be a subject to continuous developments. Constanta . motor yachts and small pleasure crafts.. SNAME.. (1995).. St. Dan Micu (ICEPRONAV Galati) for their scientific advices and for permission to publish the results regarding waterjets propulsion performances. Newport News Shipbuilding 4. Blount. 34th Wegemt School. RINA .L. (1967).. Zlatev . Venning. High Speed Propellers.. 5. Editor. 8.. Newton. Cummin.. O’Brien. London 14.I.W. Alexander.. Kruppa. Delft University of Technology.. (1986).G.L.. University of Texas 22.P. (1962). Consideratii cu privire la propulsia cu elice si propulsia cu jet a navelor rapide. Virginia 15.D Dissertation. Amoraritei.N. H. Kruppa. R. Lee. (1981). ITTC Recommended Procedures.T. LIPS JET WARTSILA 3.. The Propulsion Committee Report 23. Waterjet versus Propeller Engine Matching Characteristics. ***(2002).. Indian Institute of Technology. Totolici.J. SNAME Advance Marine Vehicles Meeting. Principle of Naval Architecture 9.. ***..E. Kuiper.J.. Editor Roy Harrington. A three day intensive course for engineers. P. Propeller Design Aspects of Large.(1992)... Wijngaarden. Propellers ’81 Symposium. C. 17.2..G.K. For high speed ships in range from 40 to 80 knots.. Hydrodynamics and Design..Morgan. ***.(1983). Cavitation. (2004). Performance Data of Propellers for High Speed Craft. ***(2002).P. With a speed range above 45 knots waterjet propulsion systems offer distinct advantages over other propulsion devices and they are applied for a wide variety of ships: fast ferries. The Propulsion Committee Report 2.I. (19). Kharagpur 19.N. High Ships. ACKNOWLEDGMENTS The author is grateful to Mr. J.Waterjet Propulsion Systems..... Micu. 18. Cavitation Descriptions and Cavitation Appearance. Gorkarn R.. Virginia 24.. Hubble.. Young. Meeting of the Royal Institution of Naval Architects 16. D. Ahn. Naval Engineering Journal 25. 6. ***(2005). ***(1987). Ph. M. P.. Kozhukharov.. (2002)... Journal of Ship&Ocean Technology.S. supercavitating propellers are considered to be the most fuel efficient propulsive.B. Kuiper.Newton Rader propellers for which the hydrodynamics performances al low cavitation numbers are known. Rader.. Testing and extrapolations methods. New Developments around Sheet and Tip Vortex Cavitation on Ships’ Propellers. Practical Aspects in the Design of high Speed Small Propellers. 79. Supercavitating Propeller Performances. Brandau. Vol... Delft University of Technology.(1997). Tachmindji.. (1962). C. Marine Screw Propellers 10. Yim. (1960). Academia Navala Mircea cel Batran. Proceeding of the 23th ITTC vol. The University of Michigam 11. REFERENCES 1. Proceeding of the 24th ITTC vol. The propeller as a source of noise and vibrations. Sizing Segmental Section Commercially Available Propellers for Small Crafts.. (1967). with higher efficiency. Proceeding of the 19en ITTC.1 20. (2000). 3rd Lips Symposium.. "Marine Engineering". G.. T. Regression Analysis of Gawn Burrill Series for Applications in Computer Aided High Speed Propeller Design. E. Kim. the tendency toward increasing the ships speeds demanding new types of propulsion devices capable to operate satisfactorily at high speeds. 12. Surface piercing propellers are recognized as an outstanding propulsion device for small high speed crafts of limited draught. Comstock. Numerical Modeling of Supercavitating and Surface-Piercing Propellers. New York. 26..(2000).

kxx [m] Roll radius of gyration If the ship moves at full speed the turning kyy [m] Pitch radius of gyration parameters are essential.obreja@ugal. In this ship operation. Ship manoeuvrability.Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics Special issue Workshop on Vortex Dominated Flows – Achievements and Open Problems Timisoara. Naval Architecture Faculty “Dunarea de Jos” University of Galati Liviu CRUDU. Romania. special viscous models in ship manoeuvrability problems. which allows the determination of understanding of the complex hydrodynamic ship’s trajectory as well as kinematical and phenomenon are necessary. simulation. Lect. the numerical and experimental analysis of for designer to use powerful hydrodynamic tools in the standard manoeuvres characteristics have been order to investigate the ship safety. environmental condition and manoeuvrability performances of the ship.course keeping (describes the performances of AWL [m2] Area of load waterline the ship to maintain the course direction). B [m] Moulded breadth .course changing (describes how fast can change D [m] Stock propeller diameter the course direction ). ∇ [m3] Volumetric displacement The occurrence of dangerous situations may have undesirable consequences for ship safety and the 1. 2007 SIMULATION OF THE SHIP STANDARD MANOEUVRES Dan OBREJA. such as: AR [m2] Rudder area . The manoeuvrability concepts NOMENCLATURE gathers distinct nautical performances [1]. Department of Naval Architecture. interest is related to the development of hydrodynamic tools in order to investigate the KEYWORDS manoeuvrability performances of the ship. In this paper. model tests The ship manoeuvrability studies the ship horizontal motions.: (+40) 236 495 400.ro [m] Draught at aft perpendicular TA ABSTRACT TF [m] Draught at fore perpendicular [s] Natural roll period Tφ To obtain a safe ship with optimum hydrodynamic performances represents a major U [Kn] Ship speed concern in the ship design and research activity. Naval Architecture Faculty “Dunarea de Jos” University of Galati Radoslav NABERGOJ.111. INTRODUCTION critical situations should be carefully investigated. Lect. Assoc. GMT [m] Transverse metacentric height . it is very important paper. starting with the carried out on a fishing vessel. It is obvious that the analysis of model is based on a set of differential equations of the dangerous situations as well as a deeper ship motions. Naval Architecture Faculty “Dunarea de Jos” University of Galati *Corresponding author: “Dunarea de Jos” University of Galati. Romania Domneasca Street. Further The ships are considered similarly to six degree of developments can be performed with potential and freedom mechanical systems. Prof. dynamical parameters of motions. Zip code: 800201. Prof.speed changing and stopping. Three main factors must be taken into account in Special interest is related to the development of new order to consider the ship hydrodynamic safety hydrodynamic tools in order to investigate the problem: hull forms.2. Assist. Still. Ocean and Environmental Engineering University of Trieste Sandita PACURARU (POPOIU). Email: dan.Tel. The mathematical initial design stage. June 1 . In order to avoid an kzz [m] Yaw radius of gyration obstacle it is indicated to change the initial course KG [m] Vertical centre of gravity direction then to try to stop the ship. no. In this respect. the LBP [m] Length between perpendiculars stopping performances are important when the ship Lmax [m] Length overall LCG [m] Longitudinal centre of gravity .

Z are the components of the external forces. while a body plan of the vessel is shown in Figure 1.42 m 0. Y.74 m2 1. N the components of the external moment. K.zG) and X. The program reliability was checked on a fishing vessel. MATHEMATICAL MODEL Taking into account ship’s motion equations related to a mobile orthogonal coordinate system Oxyz.q. Fi and M i are the external excitation forces and moments.74 m 0.65 m 0.yG.0 m 2. dt (2) where. Iyy.8 m/s U Figure 1. M.8 s kxx 2.9 m 0. typical for the Mediterranean Sea. Table 1.254 m KG GMT 0.137 m2 1.725 m LBP 25.78 m 0. in time domain. At normal speed it is important to know the course keeping performances.565 m kzz 6.205 m kyy 6.8 m 0.46 m 0. The investigations have often nautical constraints (shallow water and channels). which has been tested in the Ship Hydrodynamics Laboratories of ICEPRONAV Galati [2]. dt N (1) ∑ (M i =1 i + ri × Fi ) = ∑ ri × i =1 d (mi ⋅ vi ) .943 m 3. Body plan of the fishing vessel 2. In order to apply theoretical calculations in different design stages.0 m 0.02 m2 AR 12 Kn 1.2 s 1.667 m 296. The characteristics of the ship and of the experimental model are presented in Table 1.0 m3 0.32 m 0. but the free surface influence and the fluid viscosity are neglected. it is necessary to confirm the software validity by means of experimental tests. The main difficulty regarding simulation of the ship manoeuvres consists in determining the hydrodynamic forces applied on hull.228 m LCG 11. The theoretical manoeuvrability models determines the investigation of the motion from horizontal plane.15 m D 2 2.202 m TA 2. having the speeds vi and vector radii ri then ∑ Fi = ∑ i =1 N i =1 N N d (mi ⋅ vi ) .is slow acting. connected to the ship and using the impulse and kinetic moment theorems [3] for the masses mi. Full scale ship and model main characteristics at full loading condition Main Full Model characteristics scale scale (1/12) Lmax 32.w) is the ship’s speed in the origin of the coordinate system and ω (p. If v0 (u. The implementation of a computer code capable to simulate the ship manoeuvres in different operating conditions constitutes an important task in our research activity.r) the angular speed. then vi = v0 + ω × ri .054 m Tφ 6.171 m3 ∇ TF 2.05 m 0.083 m B 8. şi Izz represent the inertia moments then. m is the ship’s mass and Ixx. (3) Considering that the origin of the coordinate system is in the ship’s gravity centre G(xG. the system of the differential equations of motions (1) and (2) get the following simplified form ⎛ ∂u ⎞ X = m⎜ + qw − rv ⎟ ⎝ ∂t ⎠ ⎛ ∂v ⎞ Y = m⎜ + ru − pw ⎟ ⎝ ∂t ⎠ ⎞ ⎛ ∂w Z = m⎜ + pv − qu ⎟ ⎠ ⎝ ∂t (4) .575 m AWL 163.v.7 m 2.88 m 0.

the linear formulation of the motions in the horizontal plane is obtained & = mu & X e + X u u + X u& u & + Yr& r &= Ye + Yv v + Yr r + Yv& v & + rU + r &xG ) m(v (11) & + N r& r &= N e + N v v + N r r + N v& v & + mxG (v & + rU ). known also (6) ∂r ⎞ ⎛ ∂v N= I zz + mxG ⎜ + ru ⎟. (14) M ⋅s e M is the matrix of masses and inertia moments of the ship and can be always reversed. Yr& . If M −1 is the inversely of matrix M. Yr. If only first order terms are considered then the linear form is obtained. when a significant tilt angle can appear during the gyration manoeuvre. Ye. The above mentioned system can have the equivalent form ⎛ ∂u ⎞ X = m⎜ − rv − r 2 ⋅ xG + przG ⎟ ⎝ ∂t ⎠ dr dp ⎞ ⎛ ∂v Y = m⎜ + ru + xG − zG ⎟ dt dt ⎠ ⎝ ∂t K= ∂p ⎞ ⎛ ∂v I xx − mz G ⎜ + ru ⎟ ∂t ⎠ ⎝ ∂t N r& are the hydrodynamic coefficients. However. I zz r X u . being coupled with the horizontal motions. ∂t ⎝ ∂t ⎠ Mention should be made that the first equation (surge motion) is not coupled with sway and yaw ones. a straight trajectory and a constant speed of the ship (u = U) and neglecting the high order terms of the (8) and (9) equations of motions (r2 and rv). the system of the differential equations of motions becomes simplified mathematical formulation can be obtained by using Taylor’s series expansion method. The vertical motions can be neglected. a A = M −1 P B = M −1 . Yv& . neglecting the roll motion (p =0). N v . multiplying equation (14) by M −1 one obtains the equivalent form r & = As + BF (15) s e where Due to the fact that the external excitation forces and moments are multivariable functions. roll motion can be considered. (10) Noting the left side matrix M and the right one P. N r − mxGU ⎥ ⎣Nv ⎦ (13) The principal sources of the hydrodynamic excitation forces and moments are due to the propulsion and steering systems (Xe. Ne) as well as those on the submerged body due to an imposed manoeuvre of the ship (Xr. (16) . Noting the vector s = {v. characterizing the interaction of the wetted surface of the ship and the water. if the origin of the coordinate system is on the ship’s symmetry plan. mainly the horizontal motions are taken into consideration. Yv . ∂t ⎠ ⎝ ∂t (7) Furthermore. r} the coupled system of equation can be written in a matrix form (8) (9) mxG − Yr& ⎤ ⎡m − Yv& & ⎢mx − N ⎥s I N − & & v zz r ⎣ G ⎦ Yr − mU ⎤ ⎡Yv =⎢ s + Fe . In such conditions. X u& . N r . Yr . as initial conditions. then equation (13) becomes r & = P⋅s + F . Nr): X = Xe + Xr Y = Ye + Yr N = Ne + Nr . (12) ⎛ ∂u ⎞ X = m⎜ − rv − r 2 xG ⎟ ⎝ ∂t ⎠ dr ⎞ ⎛ ∂v Y = m⎜ + ru + xG ⎟ dt ⎠ ⎝ ∂t N= ∂r ⎛ ∂v ⎞ I zz + mxG ⎜ + ru ⎟ . has a simplified form & = X uu + X e (m − X u& )u & + (mxG − Yr& )r &= (m − Yv& )v Yv v + (Yr − mU )r + Ye & + (I zz − N r& )r &= (mxG − N v& )v N v v − ( N r − mxGU )r + N e . Considering. for high speed ships. ∂t K= (5) When studying ship’s manoeuvrability. N v& and as hydrodynamic derivatives. the system of the differential equations of motions (6) and (7).∂p I xx + rq (I zz − I yy ) ∂t ∂q M = I yy + pr (I zz − I zz ) ∂t ∂r N= I zz + pq (I yy − I xx ).

r . Ze) can be written in a linear form The radius of the circular trajectory can be calculated with R= U UC . v. r . r . if the ship is unstable.Matrix A characterizes the internal dynamics of the system while P represents the potential damping matrix. v. Consequently. v. v. and f3 contain the nonlinear hydrodynamic derivatives.Tejsen. The first stage of the rudder deflection is followed by a transitory domain when all parameters a time dependent and then. Yδ and (19) N v v − ( N r − mxGU )r + N e + f 3 (u . r . consider the following equations [4] One may observe that the radius R and ship’s stability parameter C are directly proportional. r . δ . Using the notations Yv v + (Yr − mU )r + Ye + f 2 (u . A bigger value of the projected longitudinal wetted surface of the ship will lead to a bigger value of N v derivative which means a better course stability of the ship. δ ) (m − X u& )u & + (mx G − Yr& )r & = f 2' (u . when the straight constant speed and no external perturbations hypothesis are considered (the homogenous form of system 14) r & = As . This leads to the utilization of nonlinear hydrodynamic models. (21) & = f 1' (u . the surface of the rudder can be increased (which means the value of N δ will &= (m − X u& )u Xe = Xδδ Yv v + (Yr − mU )r + Ye + f 2 (u . The nonlinear model of ship’s manoeuvrability. mention should be made that excessive course stability will affect the other manoeuvrability qualities of the ship (like gyration). δ ) = (24) N v v − ( N r − mxGU )r + N e + f 3 (u. then. v. which can be resolved using numerical methods: . v. v. r . Ye. The hydrodynamic excitation forces and moments due to the rudder Fe (Xe. However. by the stabilized regime when the accelerations of ship motions become zero. On the other hand. δ ) & + (mxG − Yr& )r &= (m − Yv& )v (23) Ye = Yδ δ N e = Nδ δ where. v. the nonlinear mathematical model can be reduced to a set of three first order differential equations. δ ) derivatives due to the rudder action and δ is the rudder angle. the course keeping will require continuous corrections to be applied to the steering system. δ ). r . which include the high order terms of Taylor’s series expansion of the hydrodynamic external excitation forces and moments. v. δ ) the nonlinear system of equations becomes: f 3' (u. r . In order to improve the manoeuvrability qualities. δ ) = v= δ C [(mxGU − N r )Yδ + (Yr − mU )N δ ] (N vYδ − Yv N δ ) . like turning ability. r} become Nδ are the hydrodynamic where the functions f1. v. developed by Strom . The manoeuvres at high deflections angles of the rudder require the consideration of the nonlinear hydrodynamic terms and the nonlinear inertial components. r . v. (17) s One obtains the stability criterion which can be written as C = Yv ( N r − mxGU ) + N v (mU − Yr ) > 0 . a compromise has to be found in order to finally reach general good manoeuvrability qualities. r . The course keeping stability of the ship can be performed based on the investigation of the stability of the solutions of the motion equations system. δ ) f 2' (u . Consequently. f2. = r δ ( N v Yδ − Yv N δ ) (22) increase) or the rudder deflection angle. δ ) & + (I zz − N r& )r &= (mxG − N v& )v X u u + X e + f1 (u . δ ) (m − Yv& )v & + (I zz − N r& )r & = f 3' (u . δ ) = f1' (u . r . an excessive stability will lead to a bigger radius. can be increased (up to the critical value from where the lift force and moment diminished drastically). δ ) X u u + X e + f1 (u . the coupled equations of motion (14) can be written as follows P ⋅ s + Fe = 0 (20) and the components of the vector s = {v. (mxG − N v& )v (25) r= δ C In principle. v. r . (18) The hydrodynamic derivative N v is the decisive factor on the ship’s course stability.

1 10117. in terms of hydrodynamic derivatives.40. u (t )v(t ).7 -111. MANOEUVRES SIMULATION OF THE FISHING VESSEL A simulation software for prediction of the manoeuvrability performances was implemented on Windows platform. the non-dimensional hydrodynamic forces and moments are expressed by Taylor series through a 4. In Table 2 we show the experimentally based static and dynamic hydrodynamic derivatives of the fishing vessel. using the main particulars of hull. u (t )v(t ).0 33. The . As a result of the experimental investigations.5 225. v and r for the time step t m − X u& X vv / 2 X δδ / 2 X vδ m − Yv& mxG − Yr& Yv Yvvv / 6 Yvδδ / 2 3970. δ (t )] dt dv = f 2" [t . r (t ). Experimental hydrodynamic derivatives Value G105 Derivative The numerical solutions for the unknown speeds u. Calculating the values of the speeds u(t). propeller and rudder as basic input data. r (t ).3 -6776.0 -1042. δ (t )].4 -9689.the instantaneous radius of the trajectory u 2 (t ) + v 2 (t ) R(t ) = r (t ) .5 -185.0 -278. Table 2.1 -263.4 & (t ) u (t + ∆t ) = u (t ) + ∆t ⋅ u &(t ) v(t + ∆t ) = v(t ) + ∆t ⋅ v In order to find the solutions (27) the values of the functions u(0).the trajectory of the ship when the fix coordinate system O0x0y0z0 is considered &(t ). at Froude number Fn = 0. r (t + ∆t ) = r (t ) + ∆t ⋅ r (27) mu − Yr Yrvv / 2 Yδ Yδδδ / 6 Yδvv / 2 Y0 mxG − N v& ∆t [u (t ) cosψ (t ) − v(t ) sin ψ (t )] y 0 (t + ∆t ) = y 0 (t ) + x0 (t + ∆t ) = x 0 (t ) + (29) ∆t [u (t ) sin ψ (t ) + v(t ) cosψ (t )] .6 405.7 -3111.4 -17.7 -1793.the instantaneous values of the drift angle (30) iz − N r& Nv N vvv / 6 N vδδ / 2 N r − mxG u N rvv / 2 Nδ Nδδδ / 6 Nδvv / 2 N0 β (t ) = − arctg v(t ) . The practical solutions of the mathematical models require the determination of the hydrodynamic derivatives which can be done using both theoretical and experimental methods. r (t ).2 621.4 8861.0 624.2 52.the instantaneous values of the heading angle ψ (t + ∆t ) = ψ (t ) + ∆t ⋅ r (t ) (28) .0 499. one may calculate: .4 -462.2 1085. EXPERIMENTAL MANOEUVRING DERIVATIVES In order to determine the hydrodynamic derivatives. at the full loading condition [5]. experimental static and dynamic tests with Planar Motion Mechanism (PMM) were performed on a typical Mediterranean fishing vessel model. v and r for the time step (t + ∆t ) can be obtained knowing the values of the speed u.3 4688. v(0) and r(0) for the time step t = 0 have to be known. u (t )v(t ). δ (t )] dt dr = f 3" [t .5 -4539.0 -459. dt (26) nonlinear regression procedure. U (31) The accuracy of the solutions is depending on the time step value ∆t . 3. v(t) and r(t) at each time step.du = f1" [t .4 -569.

Predicted and measured characteristics of the 10°/10° zig-zag manoeuvre S tead y turn ing ra Characteristics Advance dius Predicted 5. Port δ. selfpropelled and remote controlled model. Predicted turning circle Table 3.0 Measured 6. The IMO recommendations are fulfilled with a sufficient margin [6].25 54% The spiral test gives information regarding the course keeping performances. The turning test is a very important standard manoeuvre. Spiral test .4 12.59 Transfer / Ship Length Tactical Diameter / Ship Length Steady Diameter / Ship Length Speed Loss 1.0 Drift angle ß First Overshoot Angle [deg] Second Overshoot Angle [deg] Time to Second Execute [s] Figure 2.5 12. Zig-Zag manoeuvre 10°/10° Table 4. ψ (deg/s) 6 The zig-zag test gives information regarding the ship's response of the rudder and its yaw checking ability. Figure 4 depicts the results of spiral experimental model tests and theoretical predictions obtained with the computer program. program. Based on the experimental hydrodynamic derivatives. The predicted turning circles for 35 o rudder angle (starboard) obtained with the simulation code are depicted in Figure 2 and the calculated characteristics are shown in Table 3.simulated manoeuvres can be performed for deep. The experimental zig-zag (10°/10°) tests were performed in still water [2] with a free.4 26. calm and unrestricted water.52 3. the manoeuvrability performances of the Mediterranean fishing vessel for the standard turning circle. which are recommended by IMO.9 7. Predicted characteristics of the turning test Characteristics δ = 35° (St) Advance / Ship Length 3. in order to measure the course stability and the inertia performances.26 3. The correlation between simulation and experimental results is satisfactory.ψ(deg) First overshoot angle t(s) Second overshoot angle δ ψ (simulation) ψ (experiment) Stdb Tactical diameter Transfer Figure 3. zig-zag and spiral tests were determined. 8 . Figure 3 and Table 4 show the comparison of zig-zag results from experimental model tests and predictions with the computer 4 2 Port 0 -30 -25 -20 -15 -10 -5 -2 -4 -6 -8 0 5 10 15 20 25 30 Stbd δ (deg) Simulation Experiment Figure 4. Most common are the 10°/10° and 20°/20° tests. The diagram of the rate of change of heading with respect to rudder angle suggests that the course of the ship is stable from dynamic point of view. required by navigation authorities.

the pressures distribution can be obtained. ButterworthHeinemann 2. (2000). Crudu. G. The actual investigations will be continued considering the complete numerical approach. Chislett. further efforts have to be dedicated to this purpose to be able to reproduce the real phenomena and to provide reliable inputs in the design process. Ancona IMO Resolution A751(18) adopted on 4 November 1993. M. ICEPRONAV Galati 3. 6.. L.. Lyngby Obreja. Nabergoj. Research Project Ac 4323.Hy-7. Hover F. As a first stage the potential theory assumptions will be taken into account.5. The integration over the wetted surface of the ship will lead to the determination of the hydrodynamic coefficients as well as ship’s motions in the horizontal plane. Messina. International Symposium Technics and Technology on Fishing Vessels. Report No.. V. Bertram. limited due to the complexity of the grid generation requiring strong computer capacities in order to obtain accurate numerical solutions.Tejsen formulation. “Investigations of Hydrodynamic Safety of a Fishing Vessel”.rudder . Strom-Tejsen. J. (2001). Triantafyllou M. (2002).. The computer cod is using Abkowitz’s nonlinear mathematical model in its Strom .. (2001). During manoeuvres oblique turbulent flow occurs and phenomena like flow separations and vortexes fields have a strong influences on ship’s behaviour. “A Model Testing Technique and Method 5. Interim Standards for Ship Manoeuvrability . R. Department of Ocean Engineering.propeller will be progressively taken into account. However. The hydrodynamic derivatives have been obtained experimentally using a planar motions mechanism (PMM). Static and dynamic tests have been carried out. “Practical Ship Hydrodynamics”. In the second stage the viscous RANS methods will be considered and the interdependences between hull . Massachusetts 4. The utilization of viscous models is. (1966). Obreja D.S. “Maneuvering and Control of Marine Vehicles”. CONCLUSIONS The paper presents some results of the simulations of the standard manoeuvres for a fishing vessel. Calculating the velocity potential functions and using Bernoulli’s equation. of Analysis for the Prediction of Steering and Manoeuvring Qualities of Surface Vessells”. The correlation between the theoretical and experimental values is satisfactory but can be significantly improved. D. REFERENCES 1. for the time being.. Massachusetts Institute of Technology Cambridge. “Theoretical Formulation and Experimental Verification of Design Indexes for the Safe Operation of Mediterranean Fishing Vessel at Sea”.

2007 NUMERICAL AND EXPERIMENTAL INVESTIGATION ON A TRACTOR TUG RESISTANCE PERFORMANCE Florin PACURARU.Tel. The numerical and experimental results with and without appendages suggests a very important increase of tug resistance. action that reduce the pre-processing job even if the method accuracy is inferior to methods based on differential equations for viscous flow. The approaching way of solving the problem is based on boundary element method. The problem. INTRODUCTION Solving problems of ship hydrodynamics took a lot of attention during the last few decades. The main attention is directed to the large appendages drag problem. also known as panel method.obreja@ugal. modifications and numerical studies on the method have been carried out by many researchers.2. Xia [4] applied Dawson’s method to solve the wave problem of a ship with a φ φ0 [-] [-] velocity potential double model velocity potential perturbation velocity potential due to presence of free surface λ [m] wavelength ζ [m] wave heigh φ w [-] Fn [m] Froude number RT [kN] ship total resistance g [m/s2] acceleration of gravity . consist of potential flow aproach. The model tests have been performed at Galati University Towing Tank.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). the applications published so far have shown the significant advantage of his method in simplicity and generality. with an extended skeg is presented in this paper. Assist. Email: dan. The comparison between numerical and experimental results obtained for a Tractor Tug of about 28 m in length. Improvements. KEYWORDS ship resistance. and partly on the free surface. Numerical simulation technique of free surface fluid flow around the ship. Romania Domneasca Street.ro ABSTRACT The study of the flow around the ship represents an important step in designing an optimised hull form concerning the low resistance criterion. The presented method is widely used in hydrodynamic research field.111. Since Dawson [2] first proposed the modified Rankine Panel Method based on the Gadd [3] method linearizing the free surface condition in terms of the double-model solution. Fascicola 3. no. Assoc. Zip code: 800201.: (+40) 236 495 400. modeled by field equation (Laplace) and boundary conditions is solved by an integral method [1] in which the sources are distribuited partly on the hull. The efforts of an naval architecture are determined by decreasing ship resistance and implicit the energy consumption. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. Lect. model tests. That is why it will be discretized only the body and the free surface. only on the surface boundary. numerical methods. This is because of its importance to the design of different ships and also. generated by the large skeg. Prof. The integration is not made on the entire domain. appendages influence NOMENCLATURE ABBREVIATIONS app appendage w with w/o without 1. Naval Architecture Faculty “Dunarea de Jos” University of Galati Dan OBREJA. Romania June 1 . The prediction of the ship resistance based on the CFD methods is validated with experiment results carried out in a towing tank. the physical phenomenon of ship motion is not yet fully understood due to its complexity. Naval Architecture Faculty “Dunarea de Jos” University of Galati *Corresponding author: “Dunarea de Jos” University of Galati.

This condition is derived from Bernoulli equation. the y-axis to the direction of starboard and the z-axis pointing upwards. The potential flow assumption of inviscid and irrotational flow is made for the steady flow around the hull. The ship hull is characterized by a very small length-beam ratio and a hard chine . doublets) to arbitrarily complicated solutions. 2. Forces an moments. sinks. irrotational flow the velocity V = (u . Improvements of the technique were introduced in the project by Ni and Kim [5]. Thanks to their work non-linear calculations could be carried out for all kind of ships. MATHEMATICAL MODEL Assuming that the ship hull advances in the undisturbed water with a constant velocity U ∞ . Thus we can consider the total velocity potential as a sum of double model velocity potential and the perturbation velocity potential due to presence of free surface: (4) φ = φ0 + φ w The potential is subject to the several conditions on the hull and free-surface boundaries. which we assume that is constant. including wave resistance are computed by integrating pressure over the ship hull. The solution for this problem is to linearize the free surface boundary conditions around a known solution and solve the problem in a iterative manner. φ 02lφ wll + 2φ 0lφ 0 llφ wl + gφ wz = −φ 02lφ 0 ll on (9) 3. Upstream disturbance by a moving ship vanishes at infinity: (5) lim ∇φ = 0 r →∞ Body boundary condition requires that no fluid particle penetrate the hull surface: (6) φn = 0 Kinematic boundary condition is the mathematical formulation of the physical condition that a particle at the surface should remain at the surface all the time. He also tried to solve the fully nonlinear ship wave problems through iterations with the small perturbation approximations being employed at each iteration step. and to determine the wave and total resistance of the ship. 2 2 gζ + 0. (7) φ xζ x + φ y ζ y − φ z = 0 on z = ζ The other condition to be satisfied on the free surface comes from the fact that the pressure on the free surface must be equal to the atmospheric pressure. When a ship moves through an undisturbed free surface it generates a wave system (transverse and divergent subsystems). who also made a theoretical investigation of the damping and dispersion errors of simulation technique. Equation field and boundary conditions are expressed in terms of velocity potential. It has to be enforced by an exact mathematical expressions. w) can be expressed as a gradient of a scalar function named the velocity potential: V = ∇φ (3) Linearity of Laplace equation offers the possibility to combine elementary solutions (sources. The ship is specified in a coordinate system having its x-axis parallel to the longitudinal direction.transom stern and to treat the flow around configurations with or without a lifting surface. The objective of this work is to clarify the phenomena related to free surface flow around the SKD tractor tug hull which is moving in still water. Further improvements in stability and robustness were introduced by Janson [6]. The problem is numerically solved making use of distribution of Rankine sources on the boundary surface.5 ⋅ φ x2 + φ y + φ z2 − U ∞ = 0 on z = ζ (8) ( ) The free-surface problem described above it is difficult to solve since the free surface boundary conditions are non-linear and must be satisfied on the initially unknown wavy surface. The governing equation in potential flow method is Laplace equation (1): (1) ∇ 2φ = 0 derived continuity equation (2) for incompressible flows: ∇⋅V =0 (2) For inviscid. v. It has to be enforced by numerical trick. The incompressibility of the fluid is generally assume in all physical modelling of phenomena in which fluid is water and where the relevance of wave pressure is negligible. which contributes to the total resistance. NUMERICAL SIMULATIONS Several sets of numerical simulation were performed to analyse the hydrodynamic performances of SKD 753 tractor tug hull see Figure 1. In the first iteration the problem is linearized around the double model solution and then around the solution obtained from previous iteration. The radiation condition can not be described by an exact mathematical expression. z=0 Finally the radiation condition must be imposed (enforced) to avoid upstream waves. The coordinate system has the same speed as the ship but does not follow the ship movements such trim and sink.

shape with two knuckle.84 m 2 surface domain of 0.5 ship length upstream. The main characteristics of the ship are presented in Table 1. the calculated resistance will not be zero because of numerical error (7).831 CW Figure 2 Panel arrangement on hull Figure 3 Panel arrangement on free surface As it can be observed in diagrams 2-4. 2 ship length downstream. The authors believe that the discrepancy reasons comes from the hard chine hull shape which determine the imminence of strongly non-linear phenomena that can not be captured by the potential approach. Midship section coefficient Waterline coefficient 0. The free surface has a panel distribution of 30 panels per wave length in longitudinal direction.667 0. Figure 1 SKD 753 tractor tug lines plan Experience with the method shown that a minimum of 25-30 panels per unit length are necessary to obtain a solution which gives a good representation of wave pattern. show a good agreement for moderated Froude numbers [8]. The double body resistance should be zero since the body is symmetric and no lift is generated. Figure 6.503 m 672. where comparisons between measurements and computed results are depicted.063 m 3.78m 3 384.22 m 2 67. Figure 2.2 m 14. Figure 3. The wavelength ( λ ) of the wave generated by the ship depends on the Froude number and can be calculated by formula: λ = 2π ⋅ L ⋅ Fn2 (10) B TM LCB ∇ S S APP CB CM The number of the panels in the transverse direction has been chosen such that the aspect ratio of free surface panels to be no more than three. The hull is dicretize using 1800 panels. The calculation are done for .887 0.04 m 28. This kind of phenomena has been observed in experimental investigation.17 m 11. Table 1 Main dimensions of SKD 753 hull Main characteristics Length of waterline Length between perpendiculars Breadth of waterline Draught mean Longitudinal center of buoyancy Volumetric displacement Wetted surface (bare hull) Wetted surface appendages Block coefficient of Symbol Full scale ship LWL L BP 29.[9]. the width of the free surface is 1 ship length. The quality of the panel distribution on the linear surface potential flow can be judged by the using the double body resistance (without free surface). For higher Froude numbers the difference between computed and measured results is rising.

It can be noticed.2 0.3 Fn 0.31.28 0.35 0.15 RT measured Fn 0.2 0.39 200 150 100 50 RT computed 0 0.3 0.35 0.36 0. in figure mentioned above.25 0.3 4 6 8 2 4 6 8 Figure 2 Comparison between computed and measured resistance w app in ballast condition RT[kN] 180 160 140 120 100 80 60 40 20 0 0.2 0.32 0.26 0.2 0.3 0.31 Figure 7 depicts the pressure contours on the hull surface computed for Froude number 0.2 0.4 RT[kN] 400 350 RT measured 300 250 RT computed 200 150 100 50 0 Fn 0.3 0.3 0.3 0.24 0.15 Fn 0.3 0.25 0.38 Figure 4 Comparison between computed and measured resistance w app in ballast condition Figure 7 Comparison between pressure fields for the w and w/o appendage for Froude 0.RT[kN] 120 RT computed 100 80 60 40 20 0 0.34 0. the influence of .4 RT measured RT computed Figure 5 Comparison between computed and measured resistance w app in full load condition Figure 3 Comparison between computed and measured resistance w/o app in full load condition RT[kN] 250 RT measured Figure 6 SKD 753 flow for Froude 0.

5 0 0.5 1 1.01 -0.05 0.6 % in second.08 0.08 0.06 0.04 -0.09 0. for his valuable comments and observations during the experimental model tests. The cause of this phenomena are large appendages.02 -0. The large appendages represent a very important factor of the ship resistance increasing.04 full load no app full load app H/L 0. The influence of the appendage on ship resistance has been studied by computing two cases: hull with appendage and hull with no appendage.04 -0.5 1 1. The authors wish to thank the staff of these companies for all their support.31. 4.01 0 -0.the appendage presence on body pressure distribution by comparing the two cases: hull without appendage and hull with appendage.07 0. the Dean of the Naval Architecture Faculty from University of Galati.03 0.04 ballast no app ballast app H/L 0. the fore body lines plan with a high angle of waterplane entrance. The authors wish to express their thanks to the staff of these companies for all their support and to prof. 0.02 0.5 2 X/L Comparison between the wave-cut profiles drawn for the w and w/o appendage in full load condition .06 0.03 -0. Adrian Lungu. ACKNOWLEDGMENTS The present work has been carried out for SHIP DESIGN GROUP Galati.3 % in first figure and with 17. A comparison between appendage and no appendage case is proposed in figure 8 to show the computed free-surface spectrum drawn for Froude number 0.03 -0. The height of the stern crest rises with 13.03 0.09 0.02 0. 0.05 0. the wave pattern interference phenomena. For higher Froude numbers the experiment investigation have shown the appearance of strongly non-linear phenomena such as wave breaking and flow detachment which can not be captured by the potential methods. CONCLUDING REMARKS A numerical procedure has been employed to introspect the flow around SKD 753 hull.01 -0.5 2 X/L Comparison between wave cut profiles drawn for the w and w/o appendage in ballast condition Figure 8 Wave pattern comparison for or the w and w/o appendage Two comparisons of the wave cuts along the centre line are presented in figures 10. The comparisons have shown a good agreement with experimental results for moderate Froude numbers.07 0.11 to prove the influence of the appendage on the wave profile.5 0 0.01 0 -0.02 -0. Romania.

REFERENCES 1. Lungu, A., Raad, P., Non-linear Potential Free-Surface Flow Around the Series-60 Ship Hull, 1998 ASME Fluids Engineering Division Summer Meeting, June 21 - 25, 1998 Washington DC 2. Dawson, C., A Practical Computer Method for Solving Ship Wave Problems, 2nd International Conference on Numerical Hydrodynamics, Berkley, 1977 3. Gadd, G.E., A Method of Computing the Flow and Surface Wave Pattern Around Hull Forms, Transactions of the Royal Institute of Naval Architects, Vol. 118, 1976, pp. 207-215 4. Xia, F., Larsson, L., A calculation method for the lifting potential flow around yawed surface piercing 3--D bodies, In Proceedings of the 16'th Symposium on Naval Hydrodynamics, pp. 583--597, 1986 5. Ni,S.Y., Hifgher Order Panel Methods for Potential Flows with Linear and Non-linear Free Surface Boundary Conditions, Phd Thesis, Chalmers University of Technology, Gothenburg

6. Janson, C-E., Potential Flow Panel Methods for the Calculation of Free Surface Flows with Lift, Ph.D. Thesis, Chalmers University of Technology, 1997 7. Mierlo, K.J., Trend validation of Shipflow based on the Bare hull upright resistance of the Delft Series, Master Theses, Delft Univeristy of Technology, 2006 8. Obreja, D., Popescu, G.., Pacuraru, F., (2004) “Resistance Tests Report. Tractor Tug“, Research Project No. 344/2004, “Dunarea de Jos” University of Galati 9. Obreja, D., Pacuraru, F. Popescu, G., “Investigation of the Appendages Influence on Tractor Tug Resistance Performance”, Scientific Bulletin of the Politehnica University of Timisoara, Transactions on Mechanics, pp.129-134, 2006

Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics Special issue

Workshop on Vortex Dominated Flows – Achievements and Open Problems Timisoara, Romania, June 01 - 02, 2007

**ON THE STICK – SLIP BOUNDARY CONDITIONS AT THE WALL OF MICROCHANNELS
**

Corneliu BALAN*, Prof. Hydraulics Dept., REOROM Laboratory “Politehnica” University of Bucharest Catalin BALAN, Ph.D. Student Hydraulics Dept., REOROM Laboratory “Politehnica” University of Bucharest Diana BROBOANA, Assoc. Prof. Hydraulics Dept., REOROM Laboratory “Politehnica” University of Bucharest Roland KADAR, Ph.D. Student Hydraulics Dept., REOROM Laboratory “Politehnica” University of Bucharest

*Corresponding author: Splaiul Independentei 313, 060042, Bucharest, Romania Tel.: (+40) 21 4029705, Fax: (+40) 21 4029865, Email: balan@hydrop.pub.ro

ABSTRACT The present paper investigates the influences of slip boundary conditions on the flow in a 2D channel, at small Reynolds numbers. The numerical simulations are performed for two types of wall slip conditions: zero wall shear stress, respectively imposed slip velocity. The results evidence that changes in boundary conditions induce qualitative modifications within the flow field. Therefore, the method to alternate hydrophilic with hydrophobic walls is a promising technique for improve mixing in microchannels. KEYWORDS Channel hydrodynamics, wall slip, vorticity number, vortex, separation point NOMENCLATURE Re [-] Reynolds number Wo [-] vorticity number σw [Pa] wall shear stress vw [m/s] wall (slip) velocity 1. INTRODUCTION Multiple applications of microfluidics are today related to biofluids and their flows in channels. One major target of the Lab-on-a-Chip

studies is the control of hydrodynamics and mixing of fluids in microchannels of complex geometries, literature being abundant in studies dedicated to these subjects, for example: (i) hydrodynamic mixing in micro-bifurcations, [1], [3]; (ii) length stabilization of velocity at the entrance in a microchannel and vortical structures in contractions flows, [9]; (iii) slip at the wall of channels, [7], [12]; (iv) techniques to mix immiscible fluids at low Reynolds flow regimes in channels of different configurations, [8], [10], [13]; (v) developing direct measuring procedure of velocity field, [6]. A general review of microchannel hydrodynamics is presented in [11]. The main goal of the present study is to establish proper flow conditions to maintain an uniform distribution of cells (maximum dimension of 10 µm) in a straight channel of 800 µm width, 50 µm height and 10 mm length, for a flow rate in the range 10 µl/min to 1000 µl/min. The interest is to have no deposit and agglomeration of cells within the microchannel, therefore we have to achieve “turbulence” or to create vortical structures within the flow field. Since the maximum Reynolds number is expected to be around 100, no turbulence will be present in a straight channel, even thought the rheology of the fluid sample is non-Newtonian. Vortical structures can appear in laminar flows,

but only if the geometry is complex or/and Reynolds number is high enough or/and the fluid is non-Newtonian, see Fig. 1.

vortical structure

simulations are performed with FLUENT 6.0 code, steady Navier-Stokes solutions are obtained with coupled scheme for a Newtonian fluid with viscosity of 1 Pas and density of 1000 kg/m3. The channel is 1 mm width and 20 mm length, the upper wall being divided in segments of 1 mm long. On each segment belong to the upper wall we can consider different boundary conditions; in this paper alternate stick – slip conditions are considered. In the rest of the walls no slip boundary condition is imposed.

vortical structure

Figure 1. Vortical structures in a T-channel bifurcation (3D stream lines distribution for a shear thinning fluid at Re = 600, simulation with FLUENT code 6.0). It is well known from classical fluid mechanics that tensions in a homogeneous fluid, therefore the tensions acting on particles in suspension or at solid walls, are giving by the pressure and vorticity distributions, see [2, 5]. Therefore, we have to generate non-uniform pressure and vorticity distribution in the channel, in order to improve the mixing of cells and to avoid their deposit. There are two possibilities in the case of straight channel: (i) to introduce small rigid bodies in the flow field or to increase the roughness of the channel walls, [11, 13]; this means that we create artificially condition to generate vortical structures; (ii) to alternate at the channel wall hydrophilic (no slip boundary condition) with hydrophobic (slip boundary condition) regions, see Fig. 2 (for details about slip and adherence conditions at the walls see [10, 11]). As far as we know, the second procedure is a novel and original technique, and we intend to investigate its capabilities and applications in this paper. 2. STICK – SLIP BOUNDARY CONDITIONS The present paper is concerned with the channel flow simulations at small and medium Reynolds number. The 2D geometry is a contraction flow with aspect ratio 5:1, see Fig. 2;

Iso-velocity stream lines slip vw = 0 slip vw = 0 slip vw = 0

x - axis

v=0

slip v = 0

slip

v=0

slip

v=0

Vorticity number – Wo > 1

Figure 2. Entrance flow in a microchannel (Re = 1, 2D simulations of Newtonian flow; noslip boundary conditions with exception of the upper wall): alternate hydrophobic – hydrophilic regions on the upper wall, slip – stick (vw = 0) boundary conditions. The aim of our investigations is to put in evidence differences in the flow field due to the imposed alternate stick (adherence, vw = 0) – slip boundary conditions on the upper channel wall. The distributions of wall stress, wall pressure and vorticity number are obtained for different slip conditions. The slip conditions at the wall are of two categories: (i) imposed velocity slip (vw ≠ 0); (ii) imposed maximum admissible wall shear stress along the x - direction (in our simulations, σw = 0). In Fig. 3 are shown

. with the target to explore the magnitude of flow changes within the channel due to alternate stick – slip conditions at the wall. for different boundary conditions at the upper wall (on the other walls the adherence boundary conditions was imposed.differences at the entrance in the channel due to the imposed stick – slip constrains.e. Areas at the entrance in the channel with Wo > 1 at Re = 1. for slip condition σw = 0 and Re = 1. At this moment of our study we are not interested in establishing the relation between the type of slipping condition. without grooving or increase dramatically the roughness of the walls. respectively to create mixing between fluid layers in laminar flows. wall pressure and wall shear stress along the upper channel wall. It is well known. 2). 7 are represented the vorticity number. 4. Iso-pressure and iso-vorticity lines at the entrance in channel at Re = 1: a) no slip boundary conditions. The first aim of investigation is only qualitative. no slip (stick – vw = 0) a) stick – v = 0 slip (σw = 0) stick (vw = 0) v=0 b) slip stick stick (vw = 0) slip (vw = 0. wall materials and fluid rheology. [4].where vm in the mean velocity in the channel). The areas of vorticity number higher than one at the entrance in the channel are shown in Fig. the flow kinematics is perfect Figure 3. even at small Reynolds number. v = 0).5 vm) v=0 c) stick slip v=0 Figure 4. 6 and Fig. i. wall velocity. Simulations evidence that. b) slip (σw = 0) – stick (vw = 0) on the upper wall. 5. hydrophilic walls). the alternate stick – slip wall conditions generate areas where possible vortical structures might be present (see also Fig. With the exception of the first segment from the entrance in the channels.e. In Fig. Fig. that kinematical restriction Wo > 1 represents a necessary condition for the existence of vortices in vicinity of wall at which the fluid adheres (i. c) stick (vw = 0) – slip (vw = 0. This result is important and relevant for our application in microchannels hydrodynamics.5 vm .

2 velocity wall shear stress 5.periodic along the x – direction of the channel.013 0.007 5 slip 0.upper wall [m] Figure 8.8x10 1.0 15000 3 1. pressure wall shear stress x .5x10 4 Wall shear stress [Pa] .2x10 5 4 1.012 4 slip .σw = 0 0.0x10 3 3 3 1. therefore we do not have any detachment of boundary layer.018 Figure 7.008 0.013 0.6 pressure wall shear stress slip .0x10 2.8 0. Figure 8 discloses the wall variation of pressure and wall shear stress for the velocity slip condition vw = 0. Wall velocity and wall shear stress distribution: upper wall.0x10 1.013 5000 10000 6. even at small Reynolds number. Pressure constrain in vicinity of the separation point in planar.0x10 0. see Fig. 1 slip 0.0x10 -1. Broboana D.0x10 0.0 0.0x10 4 Wall shear stress [Pa] 15000 4 Wall velocity [m/s] 1.0 -5. 8 and Fig. Wall pressure and wall shear stress distribution: upper wall.4x10 1.2 0 0.0x10 5 3 20000 0.4x10 5 1.0x10 4. As we expected.2x10 5 2.0 0. No vortex at the entrance is obtained at this value of Reynolds number.4 0.8x10 5 Wall shear stress [Pa] Vorticity number [-] 0.015 0.upper wall [m] Figure 6.6 0.0x10 1.017 slip 0 0.2 0.016 0.8 10000 0.0 0.009 6.vw = 0.1 2. alternate slip (vw = 0..4 1.0 0. alternate slip (σw = 0) – stick (vw = 0) boundary condition. 1.upper wall [m] 2.5 vm) – stick (vw = 0) boundary condition. 7 evidences no qualitative differences Wall shear stress [Pa] Wall pressure [Pa] Figure 5. Wall shear stress takes zero values only on surfaces with imposed slip velocity (not on the surface with adherence!).014 0. 7.0x10 1.2 vorticity number wall shear stress 2.4 5000 x .011 0. The points with Wo = 0 corresponds to the perfect plug flow in the neighbourhood of the wall and a local maximum of pressure. Vorticity number and wall shear stress distribution: upper wall.6 1.0x10 8.6x10 4 4 4 Wall pressure [Pa] 1.014 slip x .014 0. [4].5x10 2.015 slip 0. alternate slip (σw = 0) – stick (vw = 0) boundary condition.018 x . between the two cases for pressure distribution and wall shear stress distribution on the segments of no slip conditions.0x10 4 4 20000 4 1.015 0.017 0..5 vmax 0.0x10 4 1.6x10 0. Wall pressure and wall shear stress distribution: upper wall.010 0. Comparison between Fig.016 0.016 3 8. steady and isochoric motion: Case I – Newtonian fluid.upper wall [m] 0. on the segments with no slip Wo ≥ 1 (with the exception of the edges) and Wo < 1 where slip velocity is present.2x10 1.0x10 5 1. It is expected that modification of symmetry in pressure and vorticity distributions across and along the channel will introduce See the paper included in this volume: Balan C. alternate slip (σw = 0) – stick (vw = 0) boundary condition.6x10 5 4 1.5 vm. One conclude from these simulations that alternate boundary conditions determine “perturbation” within the flow field.

upper wall [m] 5.upper wall [m] Figure 10. 600000 400000 WSS.009 0. Wall shear stress distribution and location of separation point.0058 x . Wall pressure [Pa] 7 7 7 7 no slip slip (σw = 0) . Wall shear stress and wall pressure distributions at the entrance in micro-channel.slip (v = vm/2) 0.e. no slip WSS.0060 separation point 0.slip vx. slip (σw = 0) .0x10 2. the effect we are looking for our particular application.5x10 2.006 0. independently on the type of boundary condition. Location of the entrance vortex and separation point on the upper wall (stream lineas distribution at Re = 200. the location of separation point is considered given by vx = 0.additional forces on the particles (cells) in suspension. for different boundary conditions. ENTRANCE VORTEX Wall shear stress [Pa] 4x10 3x10 2x10 1x10 6 no slip slip (σw = 0) . Velocity [m/s] 200000 5 . a vortex is formed immediately at the entrance in the channel. in comparison to the solution with no slip.0056 0.slip (v = vm/2) 6 Beyond the first critical value of the Reynolds number. 9. but also on the imposed boundary conditions at the upper wall.007 0. see Fig.stick (vw = 0) stick (vw = 0) .008 0.010 x . In Fig. Separation Point 6 6 0 6 -1x10 -2x10 vw ≠ 0 vw = 0 6 -3x10 6 0. The location of separation point1 is dependent not only on the value of Reynolds number (in this simulations the Reynolds number is kept constant at 200).5x10 4. see Fig.007 0. as function of different alternate stick – slip boundary conditions.0050 0.008 0. 11.0x10 4.009 0. If the stick condition is imposed on that segment (vw = 0) or the velocity wall slip is known.0052 0. The separation point is always located within the first segment of the upper wall. In the case of slip condition – σw = 0.0x10 3. the point where the velocity component at the wall changes its sign. stick – slip boundary conditions).005 0.0x10 7 7 7 Figure 9. stick .010 x .006 0.005 0.slip (σw = 0) stick (vw = 0) . for different alternate stick – slip boundary conditions.stick (vw = 0) stick (vw = 0) . the separation point is defined by the “classical condition” σw = 0. i.upper wall [m] Figure 11. 10 are represented the distributions on the upper wall of the shear stress and pressure. 3.5x10 3.stick 20 15 10 Wall shear stress [Pa] 0 -200000 -400000 -600000 0.0054 0 -5 -10 -15 -20 0.slip (σw = 0) stick (vw = 0) .

Wang Y. pp 1792-1799 4.W.. Choi C-H. (2006) The effect of elasticity on drop creation in T-shaped microchannels. Zhang J. Acad. J.R. 16. Ho C-M. 33/2006. Broboana. Calin A.. Physique 5. ACKNOWLEDGMENTS The present study was supported by the CNCSIS Romanian grant no.V. Mech.. pp 381-411 12. the cells under study can be orientated in water by an electric field).. (2005) Influence of ribbon structure rough wall on the microscale Poiseuille flow.. Meinhart C. Lee L. 127. Willaime H. Cooper-White J. Holl M. Koschwanez. D. (2004) A generating mechanism for apparent fluid slip in hydrophobic microchannels.. Balan. pp 155-160 5. experimental investigations of channel flows under microscop are needed to confirm the results of numerical simulations. 36. (2006) Effective slip and friction reduction in nanograted superhydrophobic microchannel.D.. Husny J. Stroock A.. Tom 50(64) Special Issue. Chao S-H.G. Stone H.M. without modifications of the geometry or/and roughness. Annu. (2004) Multiphase flows in microfluidics. Fluids Eng. Garcia J. Kim C-J. Berker R. Caswell B..G.C.. pp 1509-1515 13.H. The further investigations are focused to electro-rheological fluids and their flow in microchannels (in particular. Phys. The present investigations propose a novel method to control the local hydrodynamics in vicinity of the channel walls. Therefore. 18. Of course. 10404-005-0067-2 10.R. [10. (2006) Model predictive hydrodynamics regulation of microflows. Kim J. Kothare M.N..J. Arnold M. (2005) Is the vorticity number a possible quantification of a vortex?. Kang K. Microfluidics: Fluid physics at the nanoliter scale. Fluids 16. 11].R. Paris 232. The existence of wall regions with different adhesion properties for the carrier fluid might be the solution to improve the mixing in a microchannel. Sci.A. Tretheway D. pp 1140-1145 ... 4.The comparison between numerical solutions put in evidence the decreasing of vortex length when alternate stick – slip boundary conditions are imposed on the upper wall. Microeng. Rational Mech.. Vortex Hydrodynamics – Consortium REFERENCES 1. pp 155-160 7. the method to alternate hydrophilic with hydrophobic walls is a promising technique for improve mixing of fluids in microchannels. Anal.. Wang H. Meldrum D.. even at small Reynolds number (characteristic flow regime for microchannels – Re < 500).. Squires T. Koelling K. Sci.. The control and generation of vortical structures in channels is the main subject of study in micro-hydrodynamics. FINAL REMARKS The paper investigated the influences of slip boundary conditions on the flow structure in a 2D channel. Microfluid Nanofluid. (2004) Engineering flows in small devices: microfluidics toward a Lab-on-a-Chip. Jang L-S. Microfluid Nanofluid 1. Rev. Rev. pp 121-136 9. (2005) Velocity measurement in microchannels with a laser confocal microscope and particle linear image velocimetry. C. Carlson R. (1951) Sur certaines propriétés de l’effort qui s’exerce sur une paroi en contact avec un fluide visqueux.. Micromech. P.J. Bleris L. especially if the working fluids are also sensitive to the electromagnetic fields. C. pp 148-149 3. Timisoara – Trans.. 087105 8.U. Muntean T. 977-1026 11. 77. respectively (ii) vw > 0. (1967) Kinematics and stress on a surface of rest. Arch. Ulmanella U. The results evidence that.G.. Baroud C. 2005... pp 385-399 6. Non-Newtonian Fluid Mech. C. Modern Physics. The stick – slip control of the fluids at the channel wall is a scientific topic of major impact in micro and nanofluid mechanics. The numerical simulations are performed for two types of wall slip conditions: (i) σw = 0. Fluid Mech. Quake S. Ajdari A.. Balan C.. 26. different imposed boundary conditions at the wall induce qualitative changes in the whole flow field. J. J. 137.. Such studies are in progress at REOROM Laboratory from UPB.D. (2005) Microdevice end pressure evaluations with Bagley correction.R. pp 547-555 2. Bull. R.

The geometry under study is a “T” planar bifurcation with a remarkable vortex developed beyond the junction. the third derivative of pressure along the wall direction is zero in that point. 2007 PRESSURE CONSTRAIN IN VICINITY OF THE SEPARATION POINT IN PLANAR. Numerical computations of the Newtonian flow (Navier-Stokes solution at Re = 100) disclose in the neighbourhood of the separation point a particular wall pressure distribution. Ω curlv Wo := = .ro ABSTRACT The paper is concerned with the investigation of 2D flow structure in the vicinity of separation point. at any surface where the fluid adheres there is a perfect equilibrium between the vorticity magnitude and the shear strain rate. At the same time. vortex. Prof. The fluid under study is Newtonian. STEADY AND ISOCHORIC MOTION: CASE I – NEWTONIAN FLUID Corneliu BALAN*. Romania. Bucharest. the flow separation areas consists of close vortices. REOROM Laboratory “Politehnica” University of Bucharest Diana BROBOANA. the tension at the surface of adherence is related to vorticity by the following relations: (2) t = − p n + 2η 0 ω × n . Hydraulics Dept. Accordingly to the boundary layer theory. June 01 . [11]. wall NOMENCLATURE Re [-] Reynolds number Wo [-] vorticity number ρ [kg/m3] mass density η 0 . i. Fax: (+40) 21 4029865. Hydraulics Dept.02. REOROM Laboratory “Politehnica” University of Buchares *Corresponding author: Splaiul Independentei 313.pub. 060042. but our target is to extend the results to viscoelastic flows. It is of interest for our applications in biofluid mechanics to obtain a simplified model of the flow in the neighbourhood of separation point. pressure. (1) 2 D 2 trD In the very vicinity of the walls the motion is viscometric.. KEYWORDS Separation point. One main parameter of the flow kinematics is the vorticity number – Wo. The aim of the present paper is to investigate the kinematics in vicinity of separation point and to establish possible supplementary constrains for the wall stresses. If the vortex is located at the wall. which suggests that Reynolds approximation for thin films is valid in this domain. the separation point is located in the vicinity of the local maximum wall pressure and the stream line on which it lays separates the vortex from the main flow field. defined as the ratio between vorticity magnitude and stretching magnitude. The authors evidence a novel constrain for the pressure field at separation point of 2D vortices.: (+40) 21 4029705.e. defined as points at the wall where the shear stress is zero. η [Pas] zero viscosity. Assoc. in order to detect more precisely its position at the wall of blood vessels. Prof. Email: balan@hydrop. see [1] and [13]. INTRODUCTION In 2D configurations. its boundaries are established by separation (stagnation) points. viscosity function v [m/s] velocity field ω [1/s] vorticity field D [1/s] stretching Ω [1/s] spin tensor n [-] normal direction to the wall p [Pa] pressure [Pa] first normal stresses difference N1 [Pa] second normal stresses difference N2 1. Romania Tel. therefore Wo → 1. Actually. Reynolds equation vorticity.Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics Special issue Workshop on Vortex Dominated Flows – Achievements and Open Problems Timisoara..

10 y . NUMERICAL CASE The present work is based on the Navier-Stokes solution in a 2D “T” geometry with relevant separation areas.direction [m] Figure 1. 2 are disclosed the vortical structures in the planar T–bifurcation with the constant width of 25 mm. . respectively D2 (y = y**).03 0. detail with Vortex I and location of the points D1 (y = y*). Stream lines distribution in a “T” flow geometry at Re = 100 (Newtonian incompressible fluid in steady motion). 2 is considered in this study as “representative case” for the 2D vortices.0 code. 2 (segregated and coupled solutions). and t = [− p + N 1 (ω ) − N 2 (ω )]n + 2η (ω ) ω × n for viscoelastic fluids. 10 3 y* y** 4x10 5 3x10 5 WSS [Pa] 10 2 WSS dp*/dy segr. . At the end. 1 are presented for comparison the numerical steady solutions obtained with the segregate and coupled procedures.04 0. 2x10 5 1x10 5 0 10 1 dp*/dy [Pa/m] y 5 0. are shown for the Vortex I. Numerical solutions for the WSS and wall pressure gradient in the area of vortex I from the Fig.08 0. . D2. [3]. at characteristic Reynolds number of 100 (Navier-Stokes and continuity equations are solved with the coupled scheme. 2. Figure 2. where ω = ω (3) and ω = 1 2 curlv .09 -1x10 0.05 0. [9] and [14].06 0. The computations have been performed with FLUENT 6. In Fig.07 0. and the maximum wall pressure point (dp*/dy = 0). at a convergence criteria of 10-9 for velocity components and continuity equation). Since it is well known that wall shear stress (WSS) and vorticity number are vanishing at separation point. The further analysis will be focused exclusively on the Vortex I characterization and the corresponding wall stresses distributions. In Fig. a special attention in this study is given to the distribution of normal stresses at the wall where the vortex is located. coupl. a novel constrain for pressure distribution at separation point is proposed.for Newtonian fluids [2]. The separation point (WSS = 0). D1. For details about the kinematics and dynamics of vertical structures in relation to vorticity field and flow separation see [7]. The numerical case presented in Fig. One can observe major quantitative differences between solutions.

Distributions along the wall for WSS and Wo. The representation of the computed iso-pressure spectrum discloses not only the point D2 (where wall pressure takes the maximum value. see Fig.2 0. So.2 y . respectively yp/ys ≅ 0.direction Figure 3. 5 and Fig. 4 are represented the computed velocity distributions in the channel and the corresponding regions with minimum vorticity magnitude. The relative position of the pressure critic point against the separation point D1.04 0.0 0. that particular isopressure line corresponds to the separation point D1. 800 600 1.Wo [-] 400 WSS [Pa] Figure 4.07 0. the area of Vortex I and location of separation point are well defined.0 200 Separation Point .3. The regions with Wo < 0.05 0. but also the existence of a particular iso-pressure lines. coordinates is at the intersection of the tangent at the vortex I boundary with the wall). which is very close to that obtained for an oblique stagnation-point flow by Dorrepaal [4]. Velocity distributions across the channel width.8 0.4.6 0 -200 -400 -600 -800 -1000 0. D1 is also a point where pressure “diverges” at the wall. Moreover.D1 WSS Wo 1.10 vortex area y* = 60 mm 0.4 0.03 0. p = pmax). VORTEX INVESTIGATION Location of the separation point D1 is direct related to the velocity and vorticity distribution. . We remark here that the computed ratio yp/ys (the relative location of D1 against D2) equals 0. 6. The conditions for the existence of separation point at the wall of the channel are obtained numerically in Fig. p = pcr. respectively ys.06 0. Vorticity number . Figure 5. 7 (the origin of yp.09 0. see Fig. which actually separate the flow field. In Fig.396. The simulations with the coupled scheme produce results in very good agreement with the theory.1 are marked in the flow field (at D1 the vorticity magnitude is zero). 5) within the flow domain. 3 for the investigated flow. It seems that always the existence of a separation point at the wall is associated to a “saddle” pressure point (pressure critic point from Fig.08 0.

see Fig. .Figure 7. Figure 6. the iso-pressure lines p = pcr separate the flow field (p = pcr corresponds to D1 and p = pmax corresponds to D2). see Fig. and (iii) the component vx of velocity is constant up to a distance x = h from the wall. The relative position of the particular points D1 and D2. 9. one observe that (i) the iso-pressure line p = pcr is almost normal to the wall at D1. 8. i. (ii) the component vy of velocity is linear along the wall direction.e. ⎜ ⎟ ∂y ⎦ ∂y ⎝ ∂y ⎠ ⎣ ( ) (4) Figure 8. ∂ hv y ⎤ ⎡ ∂ ⎛ 3 ∂p ⎞ ⎜ ⎟ = 6η 0 ⎢ 2 v x + h ⎥. The control domain in the very vicinity of D1. the spectrum of iso-velocity vy = constant are straight lines. Exploring the kinematics details of the flow in the very vicinity of the stagnation point. This remarkable flow structure suggests that in the neighbourhood of D1 the approximation of Reynolds for the motion in thin films is valid. Spectrum of pressure.

The variation of vorticity magnitude is represented across the channel. 3). on lines of y = constant. is valid at y = y* which indicates the coordinate of the point D1 at the wall (see also Fig. Details of the flow field in the neighbourhood of separation point D1. see Langlois [8] for the mathematical expression of Reynolds equation. the vorticity diffusion flux. Relation (5) at separation point D1 is completed with a supplementary restriction for vorticity at the point D2.direction [m] Figure 10. η ∂ω .10 0.06 pressure wss y* = 60 mm 200 0 -200 -400 -600 -800 -1000 0. The distributions of normal and shear stresses along the wall from Fig. the x-direction is the normal direction to the wall.07 0. This assertion is validated by numerical computations.09 y . Additional constrains at the stagnation (separation) point and criteria of the existence of planar flow separation areas are obtained also in [4] and [6]. Figure 11. (5) ∂y 3 y = y* 22000 20000 18000 16000 pcr = 19410 Pa 800 600 400 y** = 77 mm Pressure [Pa] 14000 12000 10000 8000 6000 4000 2000 0 0. Figure 9.0125 m) the vorticity distribution takes particular values. In a 2D flow. WSS [Pa] . vorticity diffusion has one single component ( σ ) and.03 0.e. Location of the critical points D1(L284) and D2(L266) (velocity distributions superimposed on the iso-pressure lines). In vicinity of the wall (x = . 11 and Fig. Condition given by relation (5).04 0.12.08 0.0. see Fig. which confirm the existence of the critical points: (i) at D1 ω ≅ 0 . 10 confirms that relation (5) is valid at D1. In our case. ∂p = ρσ .05 0. and (ii) σ = 0 at D2. (6) σ= 0 ρ ∂n measures the rate at which vorticity is created at the boundary and diffused into the fluid [14]. the relation is equivalent at D1 with the constrain ∂3 p = 0. with p = pcr. Pressure and WSS distributions at the wall. By definition. (7) ∂y σ has to be null at the point of maximum pressure. Because the right member of (4) is constant at h = constant.in the control domain of height h from the wall. i. because at the wall the Stokes approximation is valid.

Springer. Hussain F. 285.U. Bull. pp 69-94 8. Fluid Mech.0115 -0. Sci. L280. the place of interaction between the fluid and deformable or rigid solid boundaries. simultaneously with the validity at the wall of Stokes approximation.0105 -0. pp 385-399 4. 104. In this case the numerical solutions are validated by the experiments performed in REOROM Laboratories.. ω = 0. F. Anal. Haller G.. Ma H. C. Math. Balan. R.M. Non-Newtonian Fluid Mech.. Non-Newtonian Fluid Mech.. Muntean T. (1964) Slow viscous flow. 512. L270 L285 Vorticity magnitude [1/s] D2 500 400 300 200 100 0 D1 -0. Fluid Mech. (2005) Is the vorticity number a possible quantification of a vortex?. The paper was focused to the study of stresses distribution at the wall..0125 -0. pp 148-149 3. Paris 232. Springer. ZAMP. Quart. J. Rational Mech. Timisoara – Trans. Caswell B. Lugt H.. [10] and [12]. a configuration with remarkable vortical structure at the wall. which accomplishes the classical constrain for vorticity magnitude. Sci. (2006) Viscoelastic stagnation point flow in a wake. pp 206-208 11.-D. Thompson R. (1999) A new constitutive equation and its performance in contractions flows. 138. 43. Vortex Flow Press Inc. Arch. L. New York 9.. Gersten K. Berker R. Vortex Hydrodynamics – Consortium. Naccache M. pp 707-714 6. the authors investigate the kinematics and dynamics conditions at separation point in 3D bifurcations. Heidelberg 12. (1967) Kinematics and stress on a surface of rest. relation (5). (2000) Boundary layer theory. (1951) Sur certaines propriétés de l’effort qui s’exerce sur une paroi en contact avec un fluide visqueux. L284. Macmillan. C.. (1988) On the vorticity numbers of monotonous motions. in presence of vortices.700 600 L260. Calin A. FINAL REMARKS Applications of the vortex theory are present in almost all the applied fields of fluid mechanics. R. pp 105-109 14. Canad. (1995) On the identification of a vortex. Wu J.. The result was obtained assuming that the Reynolds thin film approximation is valid in vicinity of separation point. (2004) Exact theory of unsteady separation for two-dimensional flows. (2000) Is two-dimensional oblique stagnation-point flow unique?.direction [m] Figure 12. where the locations of separation (stagnation) and re-attachment points might be also dependent on the elastic normal stresses. Berlin.. J... (1996) Introduction to vortex theory. pp 375-388 13. Anal. 86. Appl. Zhou M. 8(1). P.M. Dorrepaal J. D. The characterisation of vortical structures are of great importance especially in vicinity of the walls. Rational Mech. J. pp 257-311 7.. Jeong J. In present.J. Truesdell C...0100 ACKNOWLEDGMENTS The present study was supported by the CNCSIS Romanian grant no. Acad. Dorrepaal J. 26. Distribution of vorticity magnitude on normal direction to the wall. Balan C. pp 61-66 5. (1992) The flow of a viscoelastic fluid near a point of re-attachment. It is of interest for our further research (the computation of wall stresses in blood vessels) to test and to extend the validity of constrain (5) to viscoelastic flows. The authors obtained a supplementary condition for wall pressure at the point of separation.e. Heidelberg x . REFERENCES 1.E. Arch. pp 155160 2. L266.-Y. The hypothesis and analytical results were confirmed by the numerical simulations of a 2D Newtonian flow in a bifurcation. J. . (2006) Vorticity and vortex dynamics. Berlin. Renardy M. Tom 50(64) Special Issue. 33/2006.-Z. Schlichting H. see [5]. Broboana.0110 -0. Potomac 10. Langlois W. i. 4. Mech. Souza Mendes P.0120 -0.

RNG k. Romania. Turbulent combustion. INTRODUCTION Computational Fluid Dynamics (CFD) is efficiently used to find-out the design of original new gasoodynamical equipment. The gasodynamic geometry is axisimetricaly.* Dep. KEYWORDS Burner. of Hydraulics Technical University of Cluj-Napoca *Corresponding author: Bv Muncii103-105.Prof. RANS.: (+40) 264 410777. Dep.ε ABBREVIATIONS CFD Computational Fluid Dynamics INPUT Boundary conditions at the very entrance in the interest contour BQG Burner Quested/investigated geometry OBMR Output burner mixing region 1. and the turbulent air IN (for a requested premixed air-methane combustion rate) we have provided by imposing the swirled boundary condition (for the IN velocity). CFD investigation/prediction method was used. Dep. Prof. Assist. . and today no real progress in design can be made without using CFD. Dep. The device is energetically dimensioned up to a 3-5 m3/h CH4 flow rate (up to ~50kW). The present paper regards to the first steps/results that were performed in order to design a ‘best performing’ 50kW new swirling burner within its combustion chamber. We have done the investigations/ computations searching the maximum efficiency for an as large as possible INPUT domain –an as low as possible flow-rate potential).com ABSTRACT . of Thermal Engines and Equipments Technical University of Cluj-Napoca Florin BODE.Scientific Bulletin of the Polytechnic University of Timisoara Transactions on Mechanics Special issue Workshop on Vortex Dominated Flows – Achievements and Open Problems Timisoara. Email: victor_hodor@yahoo. Swirling. With the same objectives. Consequently –to each select overall geometry. Romania Tel. that the CFD numerical investigations might neglect the species/combustion module at the very first stage of the overall burner’s geometry prediction. Cluj-Napoca. Prof. we want to underline.of Thermal Engines and Equipments Technical University of Cluj-Napoca Paula UNGURESAN. in order to find out the way to mix air and methane gas in the proper combustion conditions. Assist. following a variety of approaches and distinct modeling strategies. 2007 CFD FIRST PREDICTION IN DESIGNING A 50KW SWIRLING BURNER WITHIN ITS COMBUSTION CHAMBER Victor HODOR. of Thermal Engines and Equipments Technical University of Cluj-Napoca Claudiu RATIU. June . As soon as we want to identify an original burner design (an as simple as possible in manufacturing requirements) once again. CFD have developed in turbulent combustion modeling. Prof. and focus our investigations on two guess overall burner geometries. we try to select two or three shapes. We first have to eliminate evident improper gasodynamic flow iINPUT constrains. 300223. Combustion chamber. CFD Gasodynamic-geometry prediction.

and then selected two geometries and some of the related computational results. NUMERICAL SIMULATIONS RESULTS We have done so far. Fig. Fig. we can see that the average velocity. First BQG –Stream Function In the Fig.1b. 2. In a first image denoted with “a” we have introduced the methane –in longitudinal direction (to draw comparisons). First BQG -velocity field By examine the related results. Second BQG –Stream Function It is to be seen that in the second gasodynamic geometry (Fig.4m/s (the methane flame speed of reaction). magnitude and/or as in-between proportion).2a and b (above).2a. is to be seen the Stream Function field –related to the same boundary conditions and to the same two BQG.proper to be revised (as position. we show the advantage becoming from the idea to introduce the methane in a radial direction to the air stream. in order to enface the benefit (we have found out).2b) the Stream Function fluctuation (alternance of the increase and the decrease of .1a and 1b below.1a. In Fig. we represent the related result –for the velocity. -the both gasodynamic geometryes seems to be proper Fig. Fig. and in the second image denoted with “b”. at the front of the OUTPUT burner region –is varying around the desired range of 0. we have to revue extra parameters/conditions –required by the conditions of a performant / complete and stabile combustion process. in order to be able to compare results for the same boundary conditions (from a minimum to a maximum flow rate limit).2a. Even that. Second BQG -velocity field by the mean of the velocity field.

Fig.3 a. we may drought out an opinion concerning the local and the overall pressure within the combustion chamber. That is why. CONCLUSIONS . as it is important in mixing process at the OBMR. it is very easy to be seen that for the first BQG.3a and b) we enface the Kinetic Energy distribution. Fig. Second BQR –preassure field the pressure losses of the flue pass of the boiler. 3. First BQR –preassure field By analyzing the pressure field in both cases. but it is to be avoided by the means of enegy consumptions and the chance to introduce instabilities. First BQG Kinetic Energy By analysing these. in the next figure (Fig. the high magnitudes of Kinetic Energy are grouped imidiatlety after the air swireled entrance.its values) in the OBMR. give us more confidence regarding the capability of ensuring the best mixing ratio –requested to ensure the proper stoichiometric ratio –around 1/10. Of great interest should be also the Kinetical energegy magnitudes.4 a.4 a.3 b. Second BQG Kinetic Energy immerging zone of the methane in the air stream –which gives us more trust in the second BQG. but for the second BQR the the KE have the highest values in the Fig. The magnitude of the pressure at OUTPUT boundary (at the exit from the combustion chamber should be greater than Fig.

New York. 2.H. Hemisphere Publ. FLUENT 6 “User’s Guide”. so the investigations might continue abandoning the first BQG. Ed.Press. B. California. ACKNOWLEDGMENTS We express our gratitude to the ‘Numerical and Parallel Computing’ Laboratory from the “Polytechnic” University of Timisoara. “Measurement and Simulation of Combustion Noise emitted from Swirl Burners with different Burner Exit Geometries”. within the frame of a national consortium research program on “Hydrodynamics of the Vortex and its Applications”. 1999.Kenneth Kuan-yun Kuo. ISBN 3-540-65373-2. ISBN-973-9404-54-5. H.Shepherd. “Mathematical Models of Turbulances”. D.Fluent Inc. 9. 11. Universität Karlsruhe. we may conclude that the confidence -the chance / probability to ensure the best combustion conditions –are in the second Burner Quested Geometry. Cluj. 2005. 2006. “Numerical simulations in Fluid Mechanics”. R. S. G.K. Romeo S. USA. Academic Press. 5.Batchelor. K. Cheng. I.J. Resiga.By cumulating the forward results. 1999. 6. M. “Numerical Simulation of Premixed Turbulent Methane Combustion”. John Wiley & Sons.. Bockhorn. Division of Combustion Technology. for the opportunity they have gives us to use FLUENT.B.. 2001. REFERENCES 1. ISBN 91-628-6443-2. Lund.H.Hodor V. 7. S. P.. Timisoara. 8. 4.. 1983. 2000. Almgren. M. B... Büchner. Lawrence Berkeley National Laboratory Berkeley. Springer. “Numerical Heat Transfer and Fluid Flow”. 10..Launder.1993 . “Computational Methods for Fluid Dynamics”. Bell. MacGrow Hill. “Principles of Combustion”.E. Lebanon. 3. 1986. Spalding. 1972. Fluent Incorporated. J..Mihaescu. Day.SeminarNotes. “Computational Aeroacustic Based on Large Eddy Simulation and Acoustic Analogies”. A.Ferziger. Pantakar.. S. Cambridge Univ. Casa Cartii de Ştiinţă. Milovan. 1980. “An Introduction to Fluid Dynamics”. “Dinamica gazelor”.

The connection between TPNB and TAD is represented in Figure 1. 5]. Fascicola 3. the accepted mathematical model of TAD consists in the equations of inviscid fluid dynamics with symmetry about the propeller axis. This approximation by convergence will be referred as the physical map T ph . etc. structural dynamics. Due to the infinite number of blades. For the time being. propeller theory NOMENCLATURE [m/s] absolute velocity c relative velocity w [m/s] transport velocity cT [m/s] Figure 1. Some authors suggest that BEM is the only model suitable for engineering applications [1]. namely an active disk. the mathematical foundation of TAD is revised involving the theory of distributions.2. INTRODUCTION The state-of-art in Horizontal Axis Wind Turbines design are the complex codes that include specialized modules for aerodynamics. 2007 3`nd Workshop on Vortex Dominated Flows Timisoara. 2007 A FOUNDATION IN DISTRIBUTIONS OF GLAUERT THEORY Theodor POPESCU. Iasi. this convergence does not have a mathematical sense.: (+40) 232 278683/2138.com version of Glauert model combined with blade element theory. KEYWORDS Glauert theory. The goal of this paper is to establish an explicit mathematical connection between the generalized mathematical model of TPNB and the generalized mathematical model of TAD. It is established an explicit mathematical connection between the generalized mathematical model of TPNB and the generalized mathematical model of TAD. { f } [-] regular distribution Subscripts and Superscripts r radial direction λ tangential direction x1 axial direction ABBREVIATIONS TAD Theory of Active Disk TPNB Theory of the Propeller with finite Number of Blades modelled as lifting lines 1. According to the terminology introduced in [4]. The main aerodynamic codes are based on BEM [2. The connection between TPNB and TAD The physical model of TAD derives directly from the physical model of TPNB by increasing the number of blades to infinite. Email: thpopescu2003@yahoo. Romania June 1 . Fluid Machines and Drives Technical University “Gheorghe Asachi” of Iasi *Corresponding author: Bvd Dimitrie Mangeron 63. Romania Tel. TAD is a simplified operational model derived from the physical model of the propeller’s theory with finite number of blades modeled as lifting lines (TPNB). In this paper. while the local load is maintained constant for each radius. The propeller with finite number of blades is transformed in a propeller with infinite number of blades. ABSTRACT Theory of active disk (TAD) is the core of the classic model of Glauert. 700050. The preferred aerodynamic model is BEM (Blade Element / Momentum) – an improved . The source model is TPNB.Scientific Bulletin of the “Politehnica” University of Timisoara Transactions on Mechanics Tom 52(66). TAD is a simplified operational model. Fax: (+40) 232 242109. The core of Glauert theory is the model of active disk. distributions. lecturer Department of Fluid Mechanics.

e2 . the local aerodynamic reactions f k are equal in norm at the same radius. The relations between C C C the densities of the singularities are similar to the lifting surface case: m k = x k ∧ f k . λ = θ = k ∈ ⎪ [ b v] ⎪ k N ⎭ ⎩ 3. P k = cT . Hence. r. x1 div{w⊗( x ∧ρc )} +{ω∧ ( x ∧ρc )} +{cO −x ∧ grad { p} + ∑ mk δ k =0 N−1 Ck ( 4) ( ) The parametric equations of the lifting lines are ⎧ ⎫ c2 ⎪ ⎪ N −1 div ⎨ρ w + p c ⎬ = ∑ P k δ k C 2 ⎪ ⎪ ⎩ ⎭ k =0 ( 5) The perturbation introduced by the lifting line “k” is modeled through the Dirac-type singulaties f k δ k . According to the ⎢ ⎣ 0 − sin θ cos θ ⎥ ⎦ physical map T ph . THE GENERALIZED MATHEMATICAL MODEL OF TPNB Let us consider the case of the free propeller with N blades. Therefore: ( 1) N−1 k =0 ∑ f kδCk The natural working space is N − 1 ⎞ ' ⎛ 3 k . we will assume an isochore motion. In the generalized theory of lifting surface.x2 .x2 .2. k = 0 r R . λ ) x1 ∈ . f k ( ) (6) ( ) ⎧( x1 . while N is increasing to infinite. the products N ⋅ f10 . by R is denoted the nonDe C .x3 Figure 2. In order to avoid the problems arising from the partial lack of associativity. The multiplication holds as a ) irregular intrinsic operation [3]. ⎢ 0⎥ k =0 Rb f ⎢ ⎣ λ⎥ ⎦ N−1 Rv ( 7) B = [ e1 . R . r. k k k . The non-inertial frame attached to the lifting line k and the field of local aerodynamic reactions f k are represented in Figure 2. we rewrite (7) as D' ( 3 . The non-inertial frame O. m k δ k and P k δ k . the restriction map is injective [4]. The property holds in the case of the lifting line. we will work in 0 0 ⎤ ⎡1 ⎢ R and A ( θ ) = ⎢ 0 cos θ sin θ ⎥ ⎥ . e3 ] is the base of the non-inertial frame ( ) attached to the blade “0”.r. THE AERODYNAMIC FORCES ON THE ACTIVE DISK Since the blades are identical.x3 ⎡ f10 ⎤ ⎢ ⎥ = ∑ ∫ B ⋅ At ( θk ) ⎢ fr0 ⎥ ϕ( 0. N ⋅ f r0 and N ⋅ f λ0 must be kept constant in (7). ⎫ ⎪ ⎪ Ck : ⎨ 2π ⎬ . R − ⎜ ⎟ U ⎜ ⎟ k =0 ⎝ ⎠ k k k inertial frame O.R . The generalized equations of TPNB are [4]: div {ρ w} = 0 2) ( N− N−1 k =0 ∑ f kδCk div{w⊗ρc} +{ω∧ρc} = −grad { p} + ∑ k=0 ⎡o⎤ ⎢ f1 ⎥ N−1 Rv ⎛ 2π ⎞⎢ o ⎥ = ∑ ∫ B ⋅ ⎜ At ( θk ) ⋅ϕ( 0. x1 . θk ) ⎟ ⎢ fr ⎥ ⎝N ⎠⎢ o ⎥ k =0 Rb ⎢f ⎥ λ ⎢ ⎣ ⎥ ⎦ ( 3) where . To see if the limit make sense mathematically.

then the products N ⋅ m1 0 remain constant too. we agree to say that S AD is an active disk instead an active annulus. mλ ( r ) = −r ⋅ f10 ( r ) (13) N ⎯⎯⎯⎯⎯⎯⎯ →∞ k =0 N ⋅ f k =const lim ∑f k δ Ck = ⎡o ⎤ ⎢ f1 ⎥ Rv 2 π ⎢o ⎥ t f r ⎥ ⋅ ϕ ( 0. Consequently. λ ∈ [ 0. In order to preserve the classic terminology. fλ = f1 = 2πr 2πr 2πr o def (8) In conclusion. 4. θk ) is a Riemann k=0 N 2π N−1 sum for t ∫ A ( θ) ⋅ ϕ( 0.θ) dθ. r. N ⋅ f r0 and N ⋅ f λ0 are 0 and maintained constant. θ ) d θ rdr ∫ ∫ B ⋅ A ( θ) ⋅ ⎢ ⎢ ⎥ Rb 0 o ⎢f ⎥ λ ⎢ ⎣ ⎥ ⎦ (9) As expected. Therefore: 0 N −1 lifting line C k : k k m1 ( r ) = r ⋅ fλ0 ( r ) .Rv ] . r . the resultant moment of equivalent aerodynamic forces on the active disk S AD is: ⎛ Rv o ⎞ M ( S AD ) = ∫ − m d σ = ⎜ ∫ − m1 2πr dr ⎟ e1 ⎜R ⎟ S AD ⎝ b ⎠ o o o o (16) (11) The comparison between (10) and (11) leads to: Rv ⎛ N −1 k ⎞ 0 = − F1 ( S AD ) = F1 ⎜ C N ⎟ ∫ f1 dr ⎜U ⎟ ⎝ k =0 ⎠ Rb The resultant of aerodynamic forces on the system of N lifting lines is: (12) ⎛ N −1 k ⎞ N −1 k M⎜ ⎜UC ⎟ ⎟ = ∑ ∫ −m ds ⎝ k =0 ⎠ k =0 Ck (17) . we define: m1 = o def 0 o def N ⋅ m0 o def N ⋅ m0 N ⋅ m1 λ r . (9) becomes: ⎛ N −1 k Tf ⎜ ⎜ ∑ f δC k ⎝ k =0 o o ⎞ = ϕ σ = δS AD f y y d f ( ) ( ) ⎟ y ⎟ ∫ ⎠ S AD lim ∑ mk δC k = (9) ⎡ o ⎤ ⎢ m1 ⎥ Rv 2 π ⎢ o ⎥ t ⎢ mr ⎥ ⋅ ϕ ( 0. r . Then. the resultant of equivalent aerodynamic forces on the active disk S AD is: ⎛ Rv o ⎞ F ( S AD ) = ∫ − f d σ = ⎜ ∫ − f1 2πr dr ⎟ e1 ⎜R ⎟ S AD ⎝ b ⎠ o o o o o This relation can be rewritten as: ⎛ N −1 k Tf ⎜ ⎜ ∑ m δC k ⎝ k =0 o o o ⎞ (15) ⎟ ⎟ = ∫ m ( y ) ϕ ( y ) d y σ = m δS AD ⎠ S AD (10) The resultant of aerodynamic forces on the system of N lifting lines is: ⎛ N −1 k ⎞ N −1 k F⎜ ⎜UC ⎟ ⎟ = ∑ ∫ − f ds ⎝ k =0 ⎠ k =0 Ck The field m ( r ) = m1 ( r ) e1 + mr ( r ) er + mλ ( r ) eλ defined in (14) represents the equivalent moment of aerodynamic reaction on the active disk. θ ) d θ rdr B A ⋅ θ ⋅ ( ) ∫∫ ⎢ o ⎥ Rb 0 ⎢m ⎥ λ ⎢ ⎣ ⎥ ⎦ The field f ( r ) = f1 ( r ) e1 + f r ( r ) er + f λ ( r ) eλ defined in (8) represents the equivalent aerodynamic reaction on the active disk. Following the path from the previous section.2π]} . fr = . we notice that T f transforms the system of N lifting lines in the annulus ( S AD ) : {( x1 .N ⋅ f10 o def N ⋅ f r0 o def N ⋅ f λ0 . the moment of local aerodynamic reactions is the same for all lifting lines. Consequently. THE AERODYNAMIC MOMENT ON THE ACTIVE DISK From (6) it derives the momentum density on the 2π We notice that ∑ At ( θk ) ⋅ϕ( 0. λ ) x1 ∈ . mr = . T f is the mathematical correspondent of T ph .r ∈ [ Rb .r. From (9). Therefore: N ⋅ mλ N −1 N ⎯⎯⎯⎯⎯⎯⎯ →∞ k =0 N ⋅ f k =const The limit from the left term will be denoted as the mathematical map T f . the propeller thrust is equal with the active disk thrust. mλ = 2πr 2πr 2πr (14) If the products N ⋅ f10 . r. mrk = 0.

R as the adjoint ) def N ⋅ P0 ( r ) 2πr f10 . Let now see how TAD transforms the singular N −1 distribution k =0 ∑ f k δCk . r. Consequently: ⎛ N −1 k Tf ⎜ ⎜ ∑ P δC k ⎝ k =0 o o ⎞ = ϕ = δS AD P y y dy P ( ) ( ) ⎟ ∫ ⎟ ⎠ S AD (TAD { f } . ) The path is suggested by (7). r . we define the ( 3 . r. ) . It can not be seen how can be mapped T f to a regular distribution or to a singular vortex distribution.R → D ) ( 3 . the axial moment of the propeller is equal with the axial moment of the active disk. R by: ) cO = −U 0 for the screw. N⋅ f r0 (20) and N ⋅ f λ0 (TAD f . it follows: 1 ⎛ ⎜ 2π ⎜ ⎝ 1 2π (22) ∫ 0 f ( x1. where cO = 0 for the turbine and ( 3 .The comparison between (16) and (17) leads to: Rv ⎛ N −1 k ⎞ 0 M1 ( S AD ) = M1 ⎜ ⎜ UC ⎟ ⎟ = − N ∫ r ⋅ fλ dr ⎝ k =0 ⎠ Rb T f is based on physical reasons. r . ϕ) = ∫ ∫ ∫ f ( x1. α ) d α ⎟ d λ ⎟ ⎠ 2π The comparison with the power of the equivalent aerodynamic forces on the active disk P ( S AD ) = − ∫ 0 f ( x1. It is necessary to extend the T f to D' map T0 : D T0 ( ϕ) = def In conclusion. it’s (18) field of application is restricted to the singular distributions concentrated on the lifting lines. λ ) 1 2π 2π 0 (21) ∫ ϕ ( x1. Hence: The map T f is the mathematical correspondent in the space D' ( 3 . θ − λ ) d θ (25) ( ) ( ) (19) Obviously. N − 1 . λ ) d λ is the first coefficient of In conclusion. it derives: 0 Pk ( r ) = cO . ϕ) = ({a0 ( f )}. α ) d α 0 2π S AD ∫ P ( y ) d yσ o (23) Renaming the variables. ϕ) = ( f . λ ) ∫ ⎞ ϕ ( x1. From (6). First. R of the physical map T ph . THE AERODYNAMIC POWER OF THE ACTIVE DISK The transport velocity formula is cT = cO + ω ∧ x . ϕ) a0 ( f ) = 1 2π 2π 0 (27) leads to: ⎛ N −1 k ⎞ P⎜ ⎜UC ⎟ ⎟ = P ( S AD ) ⎝ k =0 ⎠ (24) ∫ f ( x1. Let us ( 3 . f k + ω. λ )d λ ⋅ ∫ ϕ ( x1. 6. THE GENERALIZED MAP TAD the Fourier series in λ generated by f . then: maintained constant. mk = cO f10 − ω m1 1 2π 2π 0 ∫ ϕ( x1. r . Therefore. we derive: (TAD { f } . T0 is linear and continuous. If { f } is a regular distribution. We now define TAD : D ' map of T0 : def It follows P denote: P (r ) = o k (r ) = P (r ). R → D' ) ( 3 .T0ϕ) Let see how ∞ 2π 0 0 (26) TAD transforms regular If the products N ⋅ are distributions. r. the power of the propeller is equal with the power of the active disk. then N ⋅ P 0 remain constant too. θ − λ ) d θ d λ rdr dx1 2π 2π−λ −λ The power of the aerodynamic forces on the lifting lines is: N −1 ⎛ N −1 k ⎞ = − P⎜ C ⎟ ∑ ∫ P ds ⎜U ⎟ k =0 C k ⎝ k =0 ⎠ By the substitution θ − λ = α . 0 k = 0. 5.R . r . r.

THE VORTEX SYSTEM ASSOCIATED TO THE ACTIVE DISK We denote by k Ωb the bound vortex on the lifting Following the path from the previous section.⎛ ⎛ N −1 k ⎜ TAD ⎜ ⎜ ∑ f δC k ⎜ ⎝ k =0 ⎝ Rv N −1⎛ k =0 ⎝ N −1 k =0 ⎞ ⎞ = ⎟ ⎟. z.s ) d λdtds . ϕ ( 0.t ⎥ y. λ) dλdsdt ∫ ∫ ∫ B ⋅ A ( λ) ⋅ ⎢ 0 2π y. z.0 t ∧ y. being the vortex lines. Since dz = ( z. s ) (34) ⎞ k TAD ⎜ ⎜ ∑ Ωb δCk ⎟ ⎟ = Ωb δS AD ⎝ k =0 ⎠ o ⎛ N −1 (30) Let us notice that z.t ∧ z.t ⎥ t ∧ y. ⎜ ⎟ k ∑ AD f ⎜ ⎟= S ⎝ k=0 ⎠ ∞ Rv 2π 0 ⎤ ⎡ y1. N − 1 . by Ω kf the free vortex field generated by the lifting line C k and by S k the free vortex sheet springing from C k . r. the Let us denote . r. s ∈[ Rb .k t = y. t ⎢ ⎥ Ω0 f t 0 0 y.k t y. y 0 1 z.s = y.t 0 Rb 0 0 ⎥ y3.tλ ( λ ) ⋅ A ( λ ) ⋅ At ( λ ) ⋅ y 0 .λ = e1 ∧ y 0 and 0 .s ) = 2πr er (31) parametric equations of the axis-symmetric domain D derived through rotating S 0 around the propeller axis. T Ω δ = ⎟ ⎜ k ∑ Ω b ⎜ ⎜ ∑ Ωb δC k ⎟ ⎟ C ⎟ = = 0 0 k k ⎝ ⎠ ⎝ ⎠ Consequently. ⎢ ⎥ t ⎣ ⎦ line C k . Following the same steps as in section 3. λ ) = At ( λ ) ⋅ y ( t.t (33) Similarly. T = δ ⎟ S AD AD ⎜ ⎟ ⎜ ∑ P δC k ⎟ ⎟ = P δS AD ⎠ ⎝ k =0 ⎠ Since k 0 0 y.t ∧ y.t ⎥ ⎦ ∞ Rv We conclude that TAD is the extension of T f to D' ( 3 .0 = ∫ ∫ B ⋅ A ( θk ) ⋅ ⎢ y2. Then (28) Ωkf = Ω0 f y. we derive: ⎛ N−1 k ⎞ Ω δ ϕ T . we have N −1 k =0 0 ⎤ ⎡ y1.s .0 s . t ∈[ 0. ϕ⎟ ⎟ ⎠ ⎠ t ⎛ N −1 k TΩ ⎜ ⎜ ∑ Ωb δC k ⎝ k =0 ⎡ f10 ⎞⎢ 0 fr ⎟ ⎟⎢ ⎠⎢ 0 f ⎢ ⎣ λ ⎞ def ⎟ ⎟= ⎠ N −1 N ⎯⎯⎯⎯⎯⎯ ⎯ →∞ k =0 N ⋅ Ωb =const lim k δ k ∑ Ωb C (32) Rb ∫ B∑⎜ ⎜ TAD ∑ A ( λ ) δθk . It follows where Ωb ( r ) = o def N 0 ⋅ Ωb (r ) (e ∧ y .t ∧ z.λ .λ . we derive: We now denote z ( t . (31) shows that the bounded vorticity is conserved by TAD : the product N Ωb remains constant when N is increasing to infinite. ) ∑ Ωkf δSk 7.t ϕ( t. s. +∞ ) . We assume the following generic parametric equations for the free vortex sheet S k : (S ) : k y = y k ( t . s ) .λ = A. λ ) ( r )⎤ ⎥ ( r )⎥ dr = ⎥ ( r )⎥ ⎦ = N ∫ 2π R Rv 2π ∫ b 0 ⎡ f10 ( r ) ⎤ ⎢ ⎥ B ⋅ At ( θ) ϕ ( 0.t dsdt 0 ⎢ 0 ⎥ y.k t ∧ y. s. and for k = 0. θ) d θ ⎢ f r0 ( r ) ⎥ dr ⎢ 0 ⎥ f r ⎢ ⎣ λ ( )⎥ ⎦ ⎛ N −1 k ⎞ ⎛ N −1 k ⎞ We notice that TAD ⎜ .k t 0 y. Rv ] it follows ⎛ N −1 k ⎞ o TAD ⎜ f δ ⎟ = f δS AD ⎜∑ Ck ⎟ ⎝ k =0 ⎠ the curves of s = const. It derives that (34) are the ) ( z. ⎛ N −1 k TAD ⎜ ⎜ ∑ m δCk ⎝ k =0 ⎞ o ⎛ N −1 k ⎞ o m . Then: k rot {c } = ∑ Ωb δ k =0 N −1 C k + ∑ Ωkf δ k =0 N −1 Sk (29) k 0 Ωb We notice that ( r ) = Ωb ( r ) ⋅ er ( θk ) .t 0 Rb y ⎢ ⎣ 3.t ∧ y. t ⎢ ⎥ N Ω0 f t 0 0 ⎢ y2.R .

Hansen A. based on the above operatorial approach. we say that TAD is compatible with E ( x ) . Since the motion is isochore. The main equation of TPNB are (2) and (3).cT ) + ⎬+ rot{c} ∧ w= ∑ f kδ k (39) grad ⎨ C ρ⎭ ⎪ 2 ⎪ k=0 ρ ⎩ (36) where χ D is the distribution generated by the characteristic function of the domain D . The reason of this lack of compatibility is the momentum equation nonlinearity.s ( z. It is obviously that N Ω0 f ( t. Academiei. Finally. Ω δ = TAD ⎜ T ⎜ ∑ f Sk ⎟ ⎟ Ω⎜ ⎜ ∑ Ω f δS k ⎟ ⎟ ⎝ k =0 ⎠ ⎝ k =0 ⎠ (36) shows that the free vorticity is conserved by the map TAD . This lack of compatibility implies that the momentum equations with axial symmetry can not be accepted as the second TAD equation. Longman Scientific&Technical and John Wiley&Sons implies E (TAD x ) = 0 .0 t ∧ y.. the map TAD is compatible with the generalized mass balance equation. Moriarty P.J.c ) ⎫ p⎪ 1 ⎪ −( c. s ) 2π 0 y. . (2005) AeroDyn Theory Manual. But a0 ( f ⋅ g) = a0 ( f ) ⋅ a0 ( g) 2 + ∑ aj ( f ) ⋅ aj ( g) + bj ( f ) ⋅ bj ( g) j=1 ∞ ( ) (37) We conclude that the vortex system associated to the active disk consists in the bounded vortex annulus S AD and the free vortex domain D . j . The compatibility defined above has the following physical meaning: TAD is compatible with the balance equation transformed equation TAD ( E ( x ) ) = 0 is the same balance equation written for the motion with symmetry about the propeller axis.λ . This explains the cause of the contradictions arising in Glauert theory.. t . CONCLUSIONS TAD is a mathematical tool that allows a rigorous construction of TAD based on the source model of TPNB and a critical analysis of the hypothesis currently accepted. j = (TAD u ). s ) = o def D ∫ Ωf o z.t z. et al. NREL-2005 3. The equations (4) and (5) derives from (3).above integral is equal to where Ω f ( λ. z. Georgescu A. REFERENCES 1. TAD will be compatible with the generalized model of TPNB if TAD ( E ( x ) ) = E (TAD x ) for each equation of the model.. Consequently. Dumitrescu H. dΩ Since it follows that Ω f ( s) = − b . t . The goal of the future paper is to establish a rigorous alternative to the Prandtl correction.t ∧ z. becomes TAD u. 8.C. (1990) Aerodynamic Propeller Design (in Romanian).s ) (35) (38) It follows TAD ∑ Ωk f δ k = Ω f χD k =0 S N −1 o Therefore. Bucuresti 2. Oberguggenberger M.t ϕ( λ. 9. ds ⎛ N −1 k ⎞ ⎛ N −1 k ⎞ . Consequently. The generalized momentum balance equation (3) has the following equivalent form [4]: N−1 ⎧( c. (1992) Multiplication of Distributions and Applications to Partial Differential Equations. Ed. s ) dz . the mass balance equation (2) E ( x) = 0 if the If E ( x ) = 0 is an equation that belongs to the generalized mathematical model of TPNB. It derives: div{TAD ( w)} = 0 ( ) div{w} = 0 . from (30) and (36) it follows: o N−1 ⎛ N−1 k ⎞ o k Ω δ + Ω δ = Ω TAD ⎜ b δS +Ω f χD ⎟ k k ∑ ∑ b f AD ⎜ C S ⎟ k =0 ⎝ k=0 ⎠ We recall that TAD { f } = {a0 ( f )} . if E ( x ) = 0 . ACKNOWLEDGMENTS The author acknowledge the support from the National University Research Council grants (CNCSIS A-Consortium 33/2007). THE COMPATIBILITY BETWEEN TAD AND THE VORTEX SYSTEM ASSOCIATED TO THE ACTIVE DISK It follows that TAD is not compatible with the generalized momentum equation.

ECN-C-01-055 . *** VEWTDC . Final report (2002). Popescu Th. PhD Thesis.Verification of European Wind Turbine Design Codes. (2003) Research in the Theory of Axial Rotor with Applications to Propellers and Fans (in Romanian).4. Technical University “Gheorghe Asachi” of Iasi 5.

- Mathematical Model of Different Differential Equations
- Versteeg H K , Malalasekera W Introduction To Computational Fluid Dynamics The Finite Volume Method (Longman, 1995)(T)(267S).pdf
- Presentation
- Hull Form Hydrodynamic Optimization Based on Parametric Modeling
- Fluent Code Simulation of Flow Around a Naval Hull
- Fluent Code Simulation of Flow Around a Naval Hull
- A Study of Spin Effects on Tennis Ball Aerodynamics
- A Practical System for Hydrodynamic Optimization of Ship Hull Forms
- Linear Programming With Matlab
- A Study on the Design Optimization of an AUV by Using Computational Fluid Dynamic Analysis
- Experimental Assessment of Interference Resistance for a Series 60 Catamaran
- Wave Resistance for High-speed Catamarans
- Multiple Criteria Optimization Applied to High Speed Catamaran
- Eulers-method Best Example of Using Euler Method in Chemical Reaction
- Analysis of the Flow Around a Cruise Ferry Hull by the Means of CFD
- Mathematical Modeling and ODE
- Numerical Initial Value Problems
- 19. 4 Applications Differential Equations
- 19.2 First Order Differential Equations
- Applications of MATLAB- Ordinary Differential Equatins (ODE)
- Applications of First Order Ode's
- Analysis Techniques for Evaluating Fuel Savings
- Partial Differential Equation Models and Numerical Simulations of RNA Interaction and Gene Expression
- Lecture Notes on Numerical Analysis of Partial Differential Equations
- Differential and Difference Equations

vortex dominated Flows

vortex dominated Flows

- V. K. Vlasko-Vlasov et al- Decoration of Josephson Vortices by Pancake Vortices in Bi2Sr2CaCu2O8+delta
- E. Becu and V. Pavlov- Evolution of localized vortices in the presence of stochastic perturbations
- Piping Patent
- Thin Airfoil Theory
- C.F. Barenghi, D. Kivotides and Y.A. Sergeev- Close Approach of a Spherical Particle and a Quantised Vortex in Helium II
- 205
- GT2010-22892
- Stability Static 2
- Stem Aero Design
- 10.1.1.111
- Aerodynamics
- Design and Performance Prediction of and Axial Flow Compressor
- Unsteady Combustor Physics
- Monday
- 09_Nennemann
- flt2007_1
- Brochure SkyRanger
- 7.4 Open Channel Flow
- 00087183
- A Model for Gas-liquid Slug Flow in Horizontal and Near Horizontal Tubes
- Wall Inteference Effects - Analysis and Correction for Automotive Wind Tunnels
- 10.5923.j.mechanics.20120203.02
- SPE-90370-MS
- Acousticsby horokova
- Agitation
- Mae 252 Syllabus
- Dust Thrusters
- 2012.1_release_notes.pdf
- KeithGalvin_PhDthesis
- Kadambi 1982.pdf

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd