Está en la página 1de 8

Mathematical Basis and

Validation of the Full Cavitation
Ashok K. Singhal
e-mail: Cavitating flows entail phase change and hence very large and steep density variations in
the low pressure regions. These are also very sensitive to: (a) the formation and transport
Mahesh M. Athavale of vapor bubbles, (b) the turbulent fluctuations of pressure and velocity, and (c) the
magnitude of noncondensible gases, which are dissolved or ingested in the operating
Huiying Li liquid. The presented cavitation model accounts for all these first-order effects, and thus
is named as the ‘‘full cavitation model.’’ The phase-change rate expressions are derived
Yu Jiang from a reduced form of Rayleigh-Plesset equation for bubble dynamics. These rates de-
pend upon local flow conditions (pressure, velocities, turbulence) as well as fluid proper-
CFD Research Corporation, ties (saturation pressure, densities, and surface tension). The rate expressions employ two
Huntsville, AL 35805 empirical constants, which have been calibrated with experimental data covering a very
wide range of flow conditions, and do not require adjustments for different problems. The
model has been implemented in an advanced, commercial, general-purpose CFD code,
CFD-ACE⫹. Final validation results are presented for flows over hydrofoils, submerged
cylindrical bodies, and sharp-edged orifices. Suggestions for possible extensions of the
model implementation, e.g., to nonisothermal flows, for ingestion and mixing of noncon-
densible gases, and for predictions of noise and surface damage are outlined.
关DOI: 10.1115/1.1486223兴

Introduction veloped by the principal author and his colleagues 关10,11兴, have
had limited success, primarily due to: 共a兲 the lack of robustness of
The capability for multidimensional simulation of cavitating
numerical algorithms, and 共b兲 lack of generality of the correlations
flows is of critical importance for efficient design and perfor-
or approach used. As a result, no cavitation model was routinely
mance of many engineering devices. Some examples are: indus- used for practical CFD-based design optimization studies.
trial turbomachinery, turbopumps in rocket propulsion systems, The Full Cavitation Model described here meets all the above-
hydrofoils, marine propellers, fuel injectors, hydrostatic bearings, mentioned requirements and is already beginning to get routinely
and mechanical heart valves. In most cases, cavitation is an unde- used in industry for water and oil pumps, inducers, impellers, and
sirable phenomenon, causing significant degradation in the perfor- fuel injection systems.
mance, e.g., reduced flow rates, lower pressure increases in
pumps, load asymmetry and vibrations and noise. Multidimen-
sional simulations can enable a designer to eliminate, reduce or
shift the cavitation regions. The objective of the present study is to Description of the Full Cavitation Model
develop a practical cavitation model capable of predicting major The basic approach consists of using the standard viscous flow
performance parameters. Its extensions to prediction of cavitation 共Navier-Stokes兲 equations for variable fluid density and a conven-
related surface damage, which affects the life of the equipment, tional turbulence model 共e.g., k-␧ model兲. The fluid density is a
may be considered in future. function of vapor mass fraction f, which is computed by solving a
Numerical simulation of cavitating flows poses unique chal- transport equation coupled with the mass and momentum conser-
lenges, both in modeling of the physics and in developing robust vation equations. The ␳-f relationship is:
numerical methodology. The major difficulty arises due to the
large density changes associated with phase change. For example, 1 f 1⫺ f
the ratio of liquid to vapor density for water at room temperature ⫽ ⫹ (1)
␳ ␳v ␳l
is over 40,000. Furthermore, the location, extent and type of cavi-
tation are strongly dependent on the pressure field, which in turn and the vapor volume fraction ␣ is deduced from f as:
is influenced by the flow geometry and conditions. Therefore, in a
practical modeling approach, a priori prescription 共or assumption兲 ␳
␣⬅ f (2)
of the location and/or size of cavitation region should not be re- ␳v
quired. Likewise, the phase change correlations should have mini-
mum essential empiricism so that diverse applications can be The vapor mass fraction, f, is governed by a transport equation:
simulated without adjusting any constants or functions.

Over the last several decades, considerable effort from both ជ f 兲 ⫽ⵜ• 共 ⌫ⵜ f 兲 ⫹R e ⫺R c
共 ␳ f 兲 ⫹ⵜ• 共 ␳ V (3)
experimental and analytical fronts has been devoted to under- ⳵t
standing cavitation. For example, References 关1–12兴 include some
The source terms R e and R c denote vapor generation 共evapora-
recent reviews as well as attempts on modeling and application of
tion兲 and condensation rates, and can be functions of: flow param-
cavitation. Unfortunately, all past models, including the two de-
eters 共pressure, flow characteristic velocity兲 and fluid properties
共liquid and vapor phase densities, saturation pressure, and liquid-
Contributed by the Fluids Engineering Division for publication in the JOURNAL
OF FLUIDS ENGINEERING. Manuscript received by the Fluids Engineering Division
vapor surface tension兲.
April 20, 2001; revised manuscript received February 28, 2002. Associate Editor: J. The above formulation employs a homogenous flow approach,
Katz. also known as Equal-Velocity-Equal-Temperature 共EVET兲 ap-

Journal of Fluids Engineering Copyright © 2002 by ASME SEPTEMBER 2002, Vol. 124 Õ 617

Downloaded From: on 10/09/2014 Terms of Use:

by using the absolute value of the pres- sure difference and treating the right side as a sink term. as a first pragmatic approxima- r. all terms except ‘‘n’’ are can be derived from the generalized Rayleigh-Plesset equation as either known constants or dependent variables. For simplicity. e. in most pressure in the absence of dissolved gas. The local The present model focuses on the use of simple rational formula- far-field pressure P is taken to be the same as the cell center tions for phase change rates 共R e and R c 兲. get the following simplified equation for vapor transport: 2 Most often.. on 10/09/2014 Terms of Use: http://asme. For the objective of a practical and general model of cavitating flows. We assume that. SEPTEMBER 2002 Transactions of the ASME Downloaded From: http://asmedl. the Phase RB D 2 RB 3 DRB Dt 2 ⫹ 2 Dt 冉 冊 冉 2 ⫽ P B⫺ P ␳l ⫺ 冊 4␯l ˙ ⫺ R 2S RB B ␳ l RB (4) Change Rate expression is rewritten in terms of bubble radius. as follows: This equation provides a physical approach to introduce the ef- fects of bubble dynamics into the cavitation model. In fact. there are no general tion or ‘‘evaporation’’ rate. 共9兲 into Eq. 共12兲 is also used to model the collapse 共con- practical value. 共8兲. Eq. 共5兲. and 共10兲.org/terms . it can R e⫽ RB • ␳ 3 ␳l 冋 3 ␣ ␳ v␳ l 2 P v⫺ P 册 1/2 (13) be considered to be an equation for void propagation and. 共6兲. ‘‘n’’ and radius of bubble RB as velocity between liquid and vapor. In most Using the Rayleigh-Plesset Equation. RB . the phase change rate is propor- 2 substituting Eq.s. In a flowing liquid with zero velocity ics Formulation. 共3兲 and 共11兲. when P⬎ P B . the two- termined by the balance between aerodynamic drag and surface phase continuity equations are written as follows: Liquid phase: tension forces.兲.061We ␴ RB ⫽ (14) Vapor phase: 2 ␳ l v rel 2 ⳵ For bubbly flow regime. 共5兲–共7兲 yields a relation between the mixture density and void fraction ␣: R e ⫽C e V ch ␳␳ ␴ l v 3 ␳l 冋 2 P v⫺ P 册 1/2 共 1⫺ f 兲 (15) D␳ Dt ⫽⫺ 共 ␳ l ⫺ ␳ v 兲 D␣ Dt (8) R c ⫽C c V ch ␴ ␳ l␳ l冋2 P⫺ P v 3 ␳l 册 1/2 f (16) Here C e and C c are two empirical coefficients and V ch is a char- The vapor volume fraction ␣ can be related to the bubble number acteristic velocity. and ␳ is the mixture density. P B ⫽ P v . In Eq. A commonly used correlation in the nuclear indus- ⳵ try is 关13兴: ជ 兴 ⫽⫺R 关共 1⫺ ␣ 兲 ␳ l 兴 ⫹ⵜ• 关共 1⫺ ␣ 兲 ␳ l V (5) ⳵t 0. pressure. Unfortunately.e. the velocity slips between the liquid of RB 共important mainly during initial bubble acceleration兲. Then. hence.g. the expression tion. our primary focus is on proper account of Equation 共12兲 is referred to here as the Reduced Bubble Dynam- bubble growth and collapse. the bubble dynamics equation Phase Change Rates.. ⫽⫺ 共 ␳ l ⫺ ␳ v 兲共 n4 ␲ 兲 1/3共 3 ␣ 兲 2/3 (10) 2. Using Eqs. the generated vapor takes the form of small bubbles. 5–10% ជ 兲 ⫽R 共 ␣ ␳ v 兲 ⫹ⵜ• 共 ␣ ␳ v V (6) of liquid velocity. Combining Eqs. the typical bubble size RB is taken to be the same mixture density. In the absence of a 关1. densation兲. In the bubble flow regime. 124. this is a fairly good simplification because of the following reasons: R⫽ 共 n4 ␲ 兲 1/3共 3 ␣ 兲 2/3 ␳ 3 冋冉 ␳ v␳ l 2 P B⫺ P ␳l 2 ⫺ RB 3 D 2 RB Dt 2 冊 册 1/2 (11) 1 In most engineering devices.12兴: general model for estimation of the number density. Eq. the dependence on velocity is found/assumed to D␳ DRB be linear. RB ⳵t →0 as ␣ →0. there are plenty of nuclei for the inception cous damping. the computed flow fields strongly depend (12) upon the physical models used for the computation of local bubble where the right side of the equation represents the vapor genera- sizes and interface drag forces. and therefore the process to be different from that of the bubble growth. 618 Õ Vol. mass transport and vis- engineering situations. These relations are based on 4 the following assumptions: ␣ ⫽n ␲ RB3 (9) 3 1. V ch in Eqs. 共4兲. as a first extra computational effort in the two-fluid approach is of little approximation. the low-pressure regions. Thus. proach. Therefore. e. in most practical two-phase flow conditions. however. where cavitation occurs. the local turbulent velocity fluctuations are damping and surface tension terms 共the 2nd and 3rd term on also of this order. without the viscous turbulent flows.g. By using various limiting arguments. While such flows can be characterized by a more rigor- ous two-fluid approach. and combining Eqs. 共8兲 we obtain tional to V rel . which allows for velocity slip between the ⳵ ⳵t 共 ␳ f 兲 ⫹ⵜ• 共 ␳ f V 兲 ⫽ 共 n4 ␲ 兲 1/3共 3 ␣ 兲 2/3 ␳ v␳ l 2 P B⫺ P ␳ 3 ␳l 冋冉 冊册 1/2 liquid and vapor phases.h. 共15兲 and 共16兲 can be expressed as the for the net phase change rate R is finally obtained as: square root of local turbulent kinetic energy 冑k. of cavitation. Though we expect the bubble collapse or reliable physical models for these parameters. the following expressions for vapor generation/ ⳵ condensation rates are obtained in terms of the vapor mass frac- ជ 兲 ⫽0 共 ␳ 兲 ⫹ⵜ• 共 ␳ V (7) ⳵t tion f: where R is the net phase change rate ⫽ (R e ⫺R c ). slip between the fluid and bubbles. The bubble pressure P B is equal to the saturation vapor Bubble Dynamics Consideration. and the fact that the per unit volume phase change Mixture: rates should be proportional to the volume fractions of the donor phase. are also the regions of relatively high velocities. 共12兲. which reflects the effect of the local relative density. The relative velocity between the liquid and vapor phase is Dt Dt of the order of 1 to 10% of the mean velocity. V rel is generally fairly small. as the limiting 共maximum possible兲 bubble size. and ignoring the second-order derivative In such high-velocity regions. RB is de- To obtain an expression of the net phase change rate.. we and vapor phases are rather small.

1C e . using a probability density function 共PDF兲 ap. and flow in a Model Implementation sharp-edged orifice. 2.. the most satisfactory values were found to be R e ⫽C e 冑k ␴ ␳ l␳ v 冋 2 P v⫺ P 3 ␳l 册 1/2 共 1⫺ f v ⫺ f g 兲 (23) C e ⫽0. The Effect of Turbulence.01 seems quite Eqs. Also. estimated This practice has been found to be much simpler. or 2D axisymmetric兲. In most engineer.g. The cavitation model can be applied to any geometric sys- the turbulent pressure fluctuations as 关15兴: tem 共3D. The basis for these values is described below. C e ⫽0. The simplifications listed in items 3 and 4 above can be removed ing equipment. 5. In each case experimental data is available for The full cavitation model has been implemented into an ad. including (25) 冋 册 flows past submerged cylindrical bodies. all grid cell types 共quad. R c ⫽C e ␳ l␳ l fv (24) ␴ 3 ␳ i. Vol. The relevant features of CFD-ACE⫹ include: unstructured/ In all the simulations presented below. and thus have considerable impact on density. references 关3. ⫽ ⫹ ⫹ (19) ␳ ␳v ␳g ␳l The primary objectives of this exercise were to: completely elimi- Non-condensable gas density ␳ g is calculated as: nate negative pressure regions. over a submerged cylindrical body. Flow is assumed isothermal and fluid properties are taken as for pressure variation with time. Some points to be noted about the current cavitation model are: proach for accounting the effects of turbulent pressure fluctua- tions. Singhal et al. Even a small amount 共e. 共1兲兲 is modified as: zone兲. 10 ppm兲 of NCG can Determination of Empirical Constants C e and C c . general purpose. robust and al.39␳ k P turb (17) supported. with the consideration of the NCG effect. Good agreement has been obtained in vanced. The secondary effect can be via in. i. Numerical computations were ini- and pressure distributions. pressure-based formulation water at 300 K. 共15兲 and 共16兲 are rewritten as: binations. The recommended values of the empirical satisfactory for general use. this treat. a variety of turbulence 0. inducers. and reason- able comparison with available data and/or flow patterns. based on the operating liquid and conditions. The primary effect is due to the expansion of gas at low eral series of computations for sharp-edged orifice and hydrofoil pressures which can lead to significant values of local gas volume flows. ␳ g⫽ (20) C c ⬍C e . and also using chord cavitation兲. 关11兴 reported a for deforming/sliding domains. hex. saturation pressure of 3540 Pa and surface tension Journal of Fluids Engineering SEPTEMBER 2002. the working fluid was adaptive/hybrid grids. the operating liquid contains a finite amount of in future as outlined at the end of the paper.. no negative pressures. and 1 f v f g 1⫺ f v ⫺ f g their sensitivity to the assumed values of coefficients. The assessment criteria in- glected due to lack of a general correlation.02558 kg/m3. is prescribed as most as good as the more rigorous practice of ref. have been determined by performing sev- 关16. liquid vapor and NCG. 关11兴. with liquid and vapor densities of 1000 and for incompressible and compressible flows. In the present model.17兴.02 and 0. 50. or due to leakage or by aeration. constants C e and C c are 0. ⬘ /2兲 P v ⫽ 共 P sat⫹ P turb (18) 4. impellers and 冑k 2 P⫺ P v 1/2 axial pumps. poly兲 and arbitrary interfaces are ⬘ ⫽0. Therefore. 关18兴. C c values ␳ were varied in the range of C e to 0.01 These values were then used for many other problems. tegration of instantaneous rates in conjunction with assumed PDF 3. wide range of conditions. This approach required: 共a兲 estimation of the local values of 1. the present set of values.02 and C c ⫽0. Concurrent use of the turbulence. velocity range of operating conditions. Several experimental investiga. and nominal values were creases in the phase-change threshold pressure. grid deformation and/or and 共b兲 computations of time-averaged phase-change rates by in. commercial CFD code. 124 Õ 619 Downloaded From: http://asmedl. good convergence rates. (21) combinations of C e and C c values were tried for several orifice ␳g flow conditions 共upstream total pressure ⫽ 2. the cavitation module currently is threshold pressure value as: decoupled from heat transfer and radiation modules. Eqs. 3. and moving grids flows 共e. a finite volume. covering a wide fraction.02 and C c ⫽0. Noncondensible gas mass fraction f g is assumed to be con- stant in the flow field. arbitrary sliding interface treatment. cluded: Final Form of Full Cavitation Model. Several other postulations for slowing down the conden- RT sation 共vapor destruction兲 process were also tried. and 共b兲 Special attention to the calculated minimum pressures. CFD-ACE⫹ all cases without adjusting any coefficient values. tet. a part of the model input. 共17兲 and 共18兲. Validation of Full Cavitation Model This section presents some of the validation results for flow over a hydrofoil. The ficients.g. structures solution modules are fully supported. constant at a given temperature for the entire flow domain. This has been ne.e. models. None of these Volume fractions of NCG and liquid are modified as: were found to be very general or robust. steady-state and time-accurate tions have shown significant effect of turbulence on cavitating solution. obtain minimum pressures close to saturation pressures and obtain minimal sensitivity to pressure WP variations. non-condensable gas 共NCG兲 in dissolved state. An appropriate value of f g . discharge coef- assumed to be a mixture of ..e. The working fluid is 共a兲 Comparison of computed mass flow rates. Effect of Noncondensable Gases „NCG…. respectively. and flow pattern 共location and extent of cavitation calculation of the mixture density 共Eq. 2D planar.01. multi-media heat transfer. It was found that to reproduce experimental trends. found to be in the range 0. numerical model.01–0. The two have significant effects on the performance of the machinery on 10/09/2014 Terms of Use: http://asme. There- where the phase-change threshold pressure P v is estimated from fore.1. and 500 bar兲 and for selected hydrofoil flow cases 共representative low and ␣ l ⫽1⫺ ␣ v ⫺ ␣ g (22) high flow rates for two angles of attack. tri.. ment has been simplified by simply raising the phase-change Due to this assumption. prism. C e and C c . Both of these flows have excellent data. for leading and mid- Finally. A large number of ␣ g⫽ f g .14兴兲. All these simulations produced satisfactory results. tially performed assuming C e ⫽C c . After many hundreds of permutations and com- 冑k to replace V ch .

0. under these con- ditions. simulations were performed at ⌺⫽0. Two other grids consisting of 4250 and 14. 124. 5 and 6. ⌺Ä0.91 Static pressures on hydrofoil surface were measured at different angles of attack and Reynolds numbers. The NCG level was set to f g ⫽1 ppm. Effects of leading edge and mid-chord cavitation on the hydrodynamic forces on a hydrofoil were experimentally investigated by Shen and Dimot- akis 关19兴.700 cells were also used to check grid sensitivity of solutions. ⌺Ä1. 1. Calculated and experimental plots of Cp on the hydrofoil top surface for two of the three cases are shown in Figs. 2 Pressure variation on the suction side of a hydrofoil. 1. Velocities. A cavitation zone exists in the mid-chord region and ⌺Ä0. Calcu- lated C p values for the two higher cell count grids were found to differ less than 1%. Simulations were performed at Re⫽3⫻106 and an angle of attack of 1 deg.84. The flow tion numberÄ0. 0. and turbulence was treated using the standard k-␧ model 1 Cavitating Flow Over a Hydrofoil. Fig. 1 Computational domain and grid. working section of the hydrofoil was mounted in a water tunnel. Fig. Cavitation inception was seen at ⌺⫽0. 1.02. SEPTEMBER 2002 Transactions of the ASME Downloaded From: http://asmedl. 5 Pressure variation on the suction side of a hydrofoil.0717 N/ . 0. and grid distribution near the hydrofoil for ␣ Ä4 deg ␴ ⫽0.76 Fig. which shows the cavitation zone on the hydrofoil surface. 3 Pressure variation on the suction side of a hydrofoil. The non-dimensional pa- rameters of interest were: ␳ lU ⬁C P ⬁⫺ P v P⫺ P ⬁ Re⫽ .76. C p⫽ (26) ␮1 1 1 ␳U 2 ␳ U2 2 l ⬁ 2 l ⬁ A two-block grid consisting of 30⫻130 cells/block 共7800 cells兲 is shown in Fig. Simulations were performed at Re⫽2⫻106 and an angle of attack of 4 deg. 2 and 3 together with experimental data.91 rate was varied to change the flow Reynolds number and the angle of attack was changed by airfoil section rotation. 1. 4 Computed total volume fraction distributions at cavita- exit pressure was specified at the right 共exit兲 boundary. and good correlation is seen.38 and 0.09 was used.43. the cavitation is confined to the front of the hydrofoil.415.1 Leading Edge Cavitation. turbulence quantities and NCG mass fraction were specified at the left 共inlet兲 boundary and an Fig.8 and thickness ratio of 0. The exit pressure was varied to yield ⌺ values of 1. Calculated Cp values on hydrofoil top surface for two of the four cases are shown in Figs. mean line of 0.2 Mid-Chord Cavitation.91 and 0. 4. A typical vapor mass fraction distribution is shown in Fig. ⌺⫽ . A NACA66 共MOD兲 airfoil section with camber ratio of 0. A second-order upwind scheme was used to dis- cretize the convective fluxes.43 620 Õ on 10/09/2014 Terms of Use: http://asme. A 2-D Fig.

and the problem tests the robustness of the numerical and physical models. types of head shapes. C p⫽ (27) ␮1 0. d. 7 Volume fractions for ⌺Ä0. 9. inside the cavitation zone and dation cases presented above.9 to 2500 bars. High flow velocities near the orifice entrance generate a zone of very low pressure right after the constriction. the computed minimum pressures the recovery zone show very good agreement. Nurick 关21兴 has published extensive experimental data for cavi- tation in a sharp-edged circular orifice. where D. 10兲. P 0 . dependence on ⌺ in the cavitation regime. Calculated and experimental procedure. and the upstream total pressure. Geometrical parameters of the orifice are D/d⫽2. 0. which drive a flow through a very small orifice. Computations tal data. C d⫽ ⫽ (28) cavitation zone for ⌺⫽0. The flow was characterized clustering around the sharp-edged corner 共Fig. P b ⫽0. and ⌺⫽1. over 2000 bar. and grid distribution near a 45-degree conical fore-body Fig.5␳ 1 U ⬁2 the inlet total pressure ranging from 1. cavitation The discharge coefficient for the orifice. In all the vali- coefficients along the conical head. 1. distributions of Cp for four of the ⌺ values are shown in Fig. and 12 in. 6 Pressure variation on the suction side of a hydrofoil. 124 Õ 621 Downloaded From: http://asmedl.88 and L/d⫽5.7. showing mid-chord enon兲 and can lead to surface damage downstream of the orifice.62. while it clearly shows a square-root shown here. The experiments were conducted in a with 2800 cells 共20⫻20 cells in the first block and 20⫻120 cells water tunnel with cylindrical test objects 0. Eq. was evaluated at 0. 7. the contraction coefficient. 0.3.025 m in diameter and in the second兲 was employed to discretize the geometry with grid 0. Results were obtained cavitation model. The proper inlet pressure P ⬁ . where the upstream pres- 61⫻39 and 124⫻39 cells 共total of 7200 cells兲. show typical convergence plots for the hydrofoil and orifice cases Journal of Fluids Engineering SEPTEMBER 2002. This is a very challenging flow computation. Figures 12共a兲 and 12共b兲 similar agreements.4.7. Results for the are fairly close to the saturation pressures. Details of the body with a 45 deg conical head are cavitating flow (⌺⬎1.34 driven flow in a sharp-edged orifice is typically encountered in fuel-injectors.7). Pressure- ⌺Ä0. The discharge coefficient Cd is constant in the non- blunt heads. orifice diameter. and the C p results for the two but there were no difficulties in treating this flow with the full larger grids again differed by less than 1%.兲.5. Extensive experimental data are reported by The flow is 2-D axisymmetric. The cavitation easily The computational grid shown in Fig. This reduces the flow rate 共choking type phenom- Fig. Two other grids sures are very high. Computed results match well with the experimental data.3. 3 Cavitating Flow in a Sharp-Edged on 10/09/2014 Terms of Use: http://asme. with Re⫽ . The model correctly predicts the inception of cavitation at were performed on bodies with hemispherical. A large number of cases were computed. The NCG level was set to f g ⫽1 ppm calculated values are in very close agreement with the experimen- for the deaerated water used in the experiments. Pressure Solution and Convergence Characteristics. Experi- ments were done with a fixed exit pressure. and has received a lot of attention. C d . indicating the robustness of the numerical at ⌺⫽0. the inlet pressures and the corresponding cavitation numbers are listed in 2-D axisymmetric computational grids were built for these Table 1. 45 deg conical. 共29兲. Simulation of flows with such with 3375 and 12.3048 m in length 共1. The predicted mass flow rates from the 2800 and 5400 cell grids var- ␳ 1U ⬁d P ⬁⫺ P v P⫺ P ⬁ ied by less than 1%. Fig. A view of the P o⫺ P v m ˙ actual m˙ actual ⌺⫽ . 0. and a 2-block structured grid Rouse and McNown 关20兴. where the flow cavitates. NCG level f g was set to 15 ppm. and L are inlet diameter.0 and 1.34 is shown in Fig. because the pressure differentials involved can be very high 共up to 2500 bar兲.0 in.5␳ 1 U ⬁2 0.34. 8 has two blocks with handles the flows at very low ⌺ values. Vol. ⌺⫽ .700 cells were also used to check grid. P o⫺ P b ˙ ideal A o 冑2 ␳ 1 共 P o ⫺ P b 兲 m 2 Front Cavitation Over Submerged Cylindrical Bodies C d ⫽C c 冑⌺ (29) The present cavitation model was applied and assessed for cavi- tating flows over cylindrical submerged bodies with different where C c . and all error residuals other two cases 共hemispherical and blunt heads兲 also showed drop by at least 4 orders of magnitude. The other using the parameters defined as: grids used for grid sensitivity check had 1300 and 5400 cells. high pressure-ratios is a difficult task even for single-phase flow. 8 Computational domain and grid. respectively. and orifice . independence of the solutions. is of interest and the cavitation number ⌺ characterizes the flow: extends towards the trailing edge with decreasing ⌺. was varied to generate different flow rates. All simulations were performed at a fixed inlet U ⬁ Figure 11 shows the comparison between the predicted dis- ⫽10 m/s and exit pressure levels were varied to achieve the charge coefficients C d with Nurick’s correlation. problems.95 bar.

0 3.23兴. This work is being published separately 关22. cavitation in rocket inducers and impellers has been analyzed. These include: 1..001 1. initially while the initial condition errors in the flow are convected out after which the solution converges rapidly. Cavitation in automotive thermostatic valves. Automotive Vane and Gear pump oil pump design optimiza- tion.5 3. 11 Orifice cavitation: comparison of cavitation model pre- dictions with Nurick’s correlation 622 Õ Vol. Fig. Cavitation in rocket turbomachinery.871 1.327 1. 9 Comparison between computed and measured Cp over a fore-body with a 45-degree conical head 2.g.101 1.9 2. the basic set of equations and con- stants described in previous sections have been found to generate respectively.963 on 10/09/2014 Terms of Use: http://asme. SEPTEMBER 2002 Transactions of the ASME Downloaded From: http://asmedl.009 1. 10 Computational grid used for the sharp-edged orifice 4.446 1.590 1.0004 1. 3. Applications of the Full Cavitation Model While the results presented here focused on the validation as- pects of the cavitation model. Fig. this model has also been used suc- cessfully on a variety of different problems for research as well as commercial applications.0 ⌺ 1. e. Cavitation in diesel fuel injectors with complex multi-port geometries and time-varying geometries and pressure load- ing Table 1 Total inlet pressure and cavitation number Po ⫻105 (Pa) 1.019 1. Convergence for the orifice case shows a plateau accurate solutions with robust convergence characteristics.0 2.226 Po ⫻105 (Pa) 10 50 100 500 1000 2500 ⌺ 1.75 5.704 1. 124. and results validated against . In all of these applications.

Approximate R e .g. Acknowledgments This work was funded in part under NSF SBIR Grant No. Such details can provide a sound foundation for the de. appropriate additional equations and modules can be incorporated for the predictions of approximate location and mag. We ⫽ Weber number Likewise. Dr. Dennis Gibson of Caterpillar. „b… orifice C e . general location and approximate extents of vapor regions. condensation rates bubble size variations can also be deduced if desired. ␯ ⫽ dynamic. Vaidya and Ram Avva for their contributions in the development of two earlier models. gas volume fraction els and numerical solution procedures and computer software data ␳. ␳ v ⫽ density of mixture. d ⫽ diameter f v . f g ⫽ vapor. and in both cases the predictions from the cavitation model were in very good agreement with the experimental data. can be applied to a wide range of problems. n ⫽ bubble number density tions can be easily relaxed by solving appropriate additional trans. A V ⫽ fluid velocity vector preliminary module based on integration of Lighthill equation us. R c ⫽ vapor generation. ␣v . gas mass fraction Potential Extensions and Collaborations k ⫽ turbulence kinetic energy The current limitations in the implementation of Full Cavitation m actual ⫽ actual orifice mass flow Model in CFD-ACE⫹ include isothermal flow assumption. which laid the basis for the current . Collabo- rative efforts are encouraged to extend this model. e. This CFD code and cavitation model was applied to a number of validation and dem- onstration problems to verify the accuracy of the model and to assess the convergence performance on difficult engineering prob- lems. Inc. the model extensions mentioned above will be best ␴ ⫽ surface tension performed by universities and/or interested R&D groups working ␮. Greek nitude of cavitation induced surface damage. coupled with CFD-ACE⫹ code. ␳ e . liquid. even for very severe flow conditions involving very high pressure differentials across the on 10/09/2014 Terms of Use: http://asme. ⬘ ⫽ turbulence pressure fluctuation P turb istics such as local gradients of pressure. Q ⫽ flow rate tions. 124 Õ 623 Downloaded From: http://asmedl. C c ⫽ constants in vapor generation condensation rate expression D. The authors would like to thank: 1. ␣e . ⌺ ⫽ cavitation number Journal of Fluids Engineering SEPTEMBER 2002. density and volume frac. ␣g ⫽ liquid. The model was also applied to cavitating flow through an orifice and computed results compared well with experimental data. to include thermal effects and the prediction of cavitation damage. ␣ ⫽ angle of attack Because of the intricate inter-coupling of various physical mod. Presented here were validation results for high-speed flow cavitation on hydrofoil and submerged cylindrical bodies. 12 Convergence characteristics for two of the validation C ⫽ hydrofoil chord length cases presented above. and a m ideal ⫽ ideal orifice mass flow fixed.. kinematic viscosities in close collaboration with the authors of the present paper. RB ⫽ bubble radius mance parameters over a wide range of conditions. N. V ch ⫽ flow characteristic velocity ing a Kirchoff-Ffowacs-Williams-Hawking 共KWFH兲 solver has W ⫽ molecular weight of non-condensible gas already been developed and used on vane pump noise predictions. „a… hydrofoil. Both of these assump. P ⫽ pressure port equation. uniform mass concentration of NCG. R ⫽ universal gas constant and approximate values of turbulence intensity. U ⬁ ⫽ freestream velocity velopment of correlations for cavitation induced noise levels. DMI- 9801239. Mr. Vol. vapor structures. for his introduction and initial support for the development of cavitation model 共even though it has taken over eight years to meet the chal- lenge of developing a practical capability兲. Since the Re ⫽ Reynolds number model seems to be reasonably accurate for predictions of perfor. this support is gratefully acknowledged. and modifying corresponding parameters like P sat P v ⫽ vapor pressure and f g . and 2. The full cavitation model. vapor. and be a valuable prediction tool for design verification and optimization. The present model provides many useful flow character. Nomenclature Fig. Summary and Conclusions A comprehensive model for cavitating flows has been devel- oped and incorporated into an advanced CFD code for perfor- mance predictions of engineering equipment. it is very likely S ⫽ surface tension that the detailed flow characteristics are also in the realistic T ⫽ temperature ranges..

. H. ‘‘CFD-ACE⫹ Theory and Users’ Manuals and Tutorials. M. 2001.. 1976. ‘‘Cavitation and Pressure Distribution. 1992. ‘‘The Effect of Flow Turbulence on tensity. J. ‘‘Numerical Study of Tip Vortex Cavi.. St... and Singhal.-T. J. 124. Vaidya. R. P. Singhal.. pp. and Yamaguchi. ‘‘Application of FEDSM99-6760. M.. 1995. Y-C. Two-Phase Flow a Vertical Cylindrical Pump. S. SEPTEMBER 2002 Transactions of the ASME Downloaded From: http://asmedl. Li. E. W. E.. Li. Iowa City. New York. and Govindan. Y. 关23兴 Athavale. 194.’’ 28th AIAA Fluid Dynamics Conference. D. K. 681– 687.’’ Adv. Y.’’ ASME FED Vol.. A. 关10兴 Avva. K. ‘‘Effect of Air Injection on Vancouver.. H.. 关4兴 . K. ‘‘Numerical Analysis of 关11兴 Singhal.. A. Vol. 关21兴 Nurick. and Brennen. M. pp. ‘‘Calculation of 关16兴 Watanabe. Vancouver. A. ‘‘An Enthalpy Based HI. A. C. 1997.. Cavitation and Gas Liquid Flow in Fluid Machinery and Devices. and McNown. Honolulu.. and Gibson.. 99–106. 624 Õ Vol.. 120. ‘‘Cavitating Propeller Analysis Inside of Canada. M. 335–388. CA. B. Conf. K. CA. Fluids Eng. 2001. References ASME FED Meeting. C.’’ 关9兴 Roth. Dimensional. Cavitation Inception.. 1975. AIAA-2001- Simulation of Cavitating Flows Using a PDF Model for Phase Change.’’ Paper No. L. and Hoejijmakers. Boger.. 关18兴 CFDRC. 15–19. J. O. T. 4„3….. G. C. Ventilated Cavitation about Submerged Bodies. the Cloud Cavitation of a Hydrofoil.. Oxford University 关1兴 Kubota.. W. ‘‘Multi-Dimensional Cavitating Flows in Rocket Turbopump Elements. 98. Canada. K. FEDSM’97-3272. W. Gibeling... a on 10/09/2014 Terms of Use: http://asme. ids Eng. L... and Massah.. SC..’’ ASME FED. S.’’ Proc. 1982.’’ FEDSM99-3764.’’ ASME FED Summer Meeting. R. A. Y. Canada.. S. Stinebring. Turbulence.. Conf. 1999.. 关15兴 Hinze. H. Hilton Head. Comparison of Computational Fluid Dynamics and Experimental Results. 2000. E. D. Canada. ‘‘Prediction of Caviation Damage: A ASME J.. pp. 1994. ‘‘Numerical Analysis of One- tion. 1998. 关5兴 Hsiao. 190. 1995. K.’’ 3400. R. Vancouver. and Rott. 关17兴 Reisman. ASME Flu. Model of Cavitation. F. and Prosperetti. Hulshoff. ‘‘Shock Wave Development in the ware. and Dimotakis. M. S. H. and Brennen. E. J. ‘‘Investigations Concerning the Influence tiphase Flow. A.. H. K. A. J. Canada. R. Vancouver. pp. 关12兴 Brennen.. J. 1948. ‘‘Multi-Phase CFD Analysis of Natural and Head Forms at Zero Angle of Yaw. McGraw Hill. Kato. San Francisco.. San Francisco. H. Conf.. 22nd ATTC.. N.’’ ASME Fluids Eng. H. M. 关13兴 Markatos. 1995. ‘‘A New Modeling of Cavi.. N. and Leonard.. ASME Fluids Eng.. C. AIAA-97-1936.. 关14兴 Stoffel. ‘‘The Effect of Gas Diffusion on the Unsteady Attached Cavitation. 关20兴 Rouse. Soft- 关2兴 Wang.. Oxford. 1997. 240.. pp. K.’’ ASME FED Meeting. and Singhal. R... Johns.. Chyczewski. A. Collapse of a Cloud of Bubbles. on Hydrodynamic Forces. T. L. P. ‘‘A New Approach to Evaluate the 关19兴 Shen. A..’’ ASME FED Meeting. 关6兴 Choi.. D. Vancouver. ‘‘Orifice Cavitation and its Effect on Spray Mixing.. 1997. Y. Hilton Head Island.. M. Fluids Eng. 关8兴 Kunz. 1999. 59–96. H.’’ J. Nuclei Population Downstream of a Cavitation Zone. J. of Pressure Distribution and Cavity Length on Hydrodynamic Cavitation In- 关3兴 Keller.’’ ASME J.. 1994..’’ Iowa Institute of Hydraulic Research.. Press.. W. and Kinnas. Cavitation and Bubble Dynamics. Jiang. Paper No. ‘‘The Influence of Surface Cavitation Cavitation Erosion Power. Cavitation and Mul...’’ ASME FED Meeting..’’ ASME FED Meeting.’’ ISROMAC-8. and Singhal. D. tating Flows: A Numerical Study of Unsteady Cavitation on a Hydrofoil Sec. and Pauley. the Full Cavitation Model to Pumps and Inducers. 2nd Ed. H. A.’’ 关7兴 Fortes-Patella. 1997.’’ 关22兴 Athavale. tation Inception Using a Bubble Dynamics Model. 1989. H. Fluid Mech. Eng. and Reboud. K. Duttweiler. H. C. 1997. A. and Schuller.